T
HE EFFECTS OF
STI
RISK COMMUNICATION
The effects of STI risk communication
How does the manipulation of STI probability and severe STI consequences
affect the perceived STI susceptibility and perceived STI severity of students in
higher education?
Evi A. Boits – 11401435 Master’s Thesis
Graduate School of Communication University of Amsterdam
Master’s programme Communication Science Thesis supervisor: Dhr. dr. G.J. de Bruijn
Abstract
Students in higher education lack in perceiving the risk of contracting sexually transmitted infections (STIs) as a result of unprotected sexual intercourse before both sexual partners are tested on STIs. This study examined how the manipulation of STI probability and severe STI consequences can heighten their STI risk perceptions: perceived STI susceptibility and perceived severity. The focus lies on the underlying processes of the manipulation of STI risk on the separate risk perceptions. 158 students participated in the online survey-embedded experiment and were randomly allocated to one out of four conditions. Results indicated that the message evaluations (i.e. readability, personal relevance and novelty) and message induced emotions (fear, worry, and satisfaction) have a cofounding role in the relationship between communicating STI risk and the related STI risk perceptions. Furthermore, presenting high STI probability information and low severe consequences successfully increases perceived STI susceptibility. Both low and high severe STI consequences were on average perceived as equally severe. Additionally, no significant effects emerged for the manipulation of severe STI consequences on neither one of the risk perceptions, nor for the interaction effect. To conclude, perceiving risk is a complex cognitive and affective process. How to optimally manipulate the constructs of risk and examine the effects on the separate risk perceptions remains a vital topic for future research.
Keywords: sexually transmitted infections, risk communication, STI probability,
severe STI consequences, risk perceptions, perceived STI susceptibility, perceived STI severity, message evaluation, message induced emotions
Introduction
Each day over one million people worldwide are contaminated with sexually
transmitted infections (STIs) (WHO, 2017). STIs are often transmitted through unprotected sexual behaviour, i.e. no condom use before both sexual partners are tested on STIs (Soa Aids Nederland, 2017a). Contamination with an STI can result in a variety of diseases categorised as sexually transmitted diseases (STDs) (WHO, 2017). Although some of these STDs, such as Chlamydia, are easily treated when detected, others such as HIV might cause permanent health damage (Aral, 2001; Soa Aids Nederland, 2017a).
Especially students in higher education (i.e. university and college students) are at large risk for contracting STIs due to their risky sexual behaviour (Downing-Matibag & Geisinger, 2009; RIVM, 2017). They engage for example in more casual relationships with often multiple sexual partners and inconsistently use of condoms (CDC, 2016; Downing-Matibag & Geisinger, 2009). The consequences of inconsistent condom use are not only STIs but also unwanted pregnancies (Hood, Hogben, Chartier, Bolan, & Bauer, 2014). However, health statistics show that students do attempt to prevent unwanted pregnancies through the use of hormonal contraceptive methods such as the birth control pill and intrauterine device (IUD) (Moscou, 2016; Soa Aids Nederland, 2017b). Although these hormonal contraceptive methods are effective in preventing unwanted pregnancies, they do not protect against STIs (Cates Jr & Steiner, 2002; Hood et al., 2014). Therefore, correct and consistent condom use remains one of the most effective strategies for preventing STIs before both sexual partners are tested (CDC, 2016; Crosby & Bounse, 2012).
Nevertheless, condom use appears to be a rather challenging behaviour influenced by many factors; from preparatory behaviours such as buying and carrying condoms (van Empelen & Kok, 2006) to the necessary skills to use them correctly (Witte, Berkowitz, Cameron, & McKeon, 1998). One of the principal factors influencing students’ decision to
not use condoms is their lack in perceiving the possible consequences of not using condoms; namely the risk for contracting an STI (Downing-Matibag & Geisinger, 2009). Not perceiving STI risk implies that students do not feel vulnerable for contracting STIs and have varying attitudes towards the severity of infection (Barth, Cook, Downs, Switzer, & Fischhoff, 2002; Newby, Wallace, & French, 2012; Wolfers, de Zwart, & Kok, 2011). For example, a study among adolescents in the Netherlands showed that most of the sexually active adolescents tend to underestimate their risk of contracting an STI (Wolfers et al., 2011). Even more worrying is that students indicate that most STIs (except for HIV and Aids) do not have severe consequences for one's health (Barth et al., 2002; Newby et al., 2012). Increasing students STI risk perceptions is, therefore, an essential and necessary component to motivate them towards condom use (Sheeran, Harris, & Epton, 2014).
A strategy to heighten the risk perceptions is communication risk (Sheeran et al., 2014). Many studies and interventions tried to target student’s perceived misconceptions about their vulnerability and the seriousness of contracting an STI (e.g., Mevissen, Ruiter, Meertens, & Schaalma, 2010; Mevissen, Ruiter, Meertens, Zimbile, & Schaalma, 2011; and Nan, Dahlstrom, Richards, & Rangarajan, 2015). In theory, risk communication is commonly studied as a part of social-psychological models such as Protection Motivation Theory (PMT) (Floyd, Prentice-Dunn, & Rogers, 2000) and the Health Belief Model (HBM) (Rosenstock, Strecher, & Becker, 1988). These models are all based on the assumption that communicating risk will increase an individual’s risk perceptions and will motivate them towards protective behaviour (Lipkus, 2007; Mevissen, Meertens, Ruiter, Feenstra, & Schaalma, 2009; Sheeran et al., 2014), for example condom use.
Although risk communication strategies are crucial elements in the social-psychological models (Sheeran et al., 2014), it is essential to make a clear conceptual
when examining how to optimally communicate risk to increase the related risk perceptions. The theoretical framework will elaborate more in detail on the importance of this distinction.
The mixture of communication strategies causes a lack of focus on the underlying processes of risk communication. This makes it rather difficult to draw conclusions about the effectiveness and best strategies for risk communication.
Furthermore, Montanaro and Bryan (2014) argue that not all constructs of social-psychological models have an equally strong effect on the outcome variables. Hence, for risk communication focussing on the two risk constructs (‘probability’ and ‘severity of the consequences’ (Schmälzle, Renner, & Schupp, 2017, p. 164)), can be more efficient than manipulating all risk constructs (Montanaro & Bryan, 2014).Therefore it is vital to examine how the manipulation of the individual constructs of risk affect the separate risk perceptions: perceived severity and perceived susceptibility. The emphasis lies on the separate risk perceptions because perceiving them as one assembled variable proves to be inaccurate and often low in reliability (Moyer-Gusé, Chung, & Jain, 2011; Witte & Allen, 2000). Adding to that, risk perceptions are specific to the health threat they represent (Ferrer & Klein, 2015; Fischhoff, Bostrom, & Quadrel, 1993). Therefore is specific examination of how the
manipulation of STI risk affects the perceived STI susceptibility and perceived STI severity of students in higher education of great relevance. This leads to the following main question of the study: ‘How does the manipulation of STI probability and the severe STI consequences affect the perceived STI susceptibility and perceived STI severity of students in higher
Theoretical Framework STI Risk Communication
Before the elaboration on STI risk and how the manipulation can affect the separate STI risk perceptions, the problem with risk communication in recent theories will be
discussed further, as it will contribute to the better understanding of the purpose of this study. As argued, risk communication is a critical determinant in numerous
social-psychological models, such as HBM (Rosenstock et al., 1988), PMT (Floyd et al., 2000), and the Extended Parallel Process Model (EPPM) (Witte, 1992). For example, HBM states that individuals are motivated through an increase in the following five constructs ‘perceived susceptibility’, ‘perceived severity’, ‘perceived benefits’, ‘perceived barriers’ and ‘self-efficacy’ (Montanaro & Bryan, 2014). Other models, such as EPPM and PMT have a more step-by-step motivational approach towards preventive behaviour. The first step of the process towards preventive behaviour focuses on the realisation of the presented risk (i.e. perceived susceptibility and perceived severity). The second is the selection of an appropriate coping strategy to deal with the perceived risk (Floyd et al., 2000; Witte, 1992). To conclude, these models agree that heightening risk perceptions is essential to motivate individuals towards protective sexual behaviour (Sheeran et al., 2014).
Although risk communication plays an essential role in these social-psychological models, they are not inherently connected to each other. To illustrate: the conceptualisation of risk is often strongly intertwined with the concept of fear and the related fear appeal studies, of which communicating risk is an important aspect (e.g. EPPM (Witte, 1992)). Although the two concepts are ‘related’, they are also ‘conceptually distinct’ (Witte & Allen, 2000, p. 592). Fear relies on affective processing and has the main goal to elicit an emotion, while risk relies on cognitive processing and has the main goal to increase the perceived risk perceptions (Witte & Allen, 2000). Because the focus of this study is to gain more insight into the
cognitive processing of the manipulated risk, the distinction between fear and risk is necessary.
A second consequence of the strong association between risk communication and the social-psychological models is that the effectiveness of risk communication is often judged based on the result of the recommended preventive behaviour (Lipkus, 2007). Moreover, although some studies, as illustrated in the meta-analysis of Sheeran et al. (2014), argue that safe sexual intentions and behaviour can be affected by communicating STI risk and the heightened risk perceptions. Others argue that increasing risk perception does not directly lead to behaviour change, as the relationship between risk communication and safe sexual intentions and behaviour can still be affected by numerous other variables (Lipkus, 2007; Montanaro & Bryan, 2014). Although the discussion about the possible gap between intentions and behaviour is not the focus of this study, it does underline the importance of preciseness in defining the intended outcome variables.
In this study, risk will be defined based on two of its core components ‘the probability of a hazardous event and the severity of the negative consequences’ (Schmälzle, Renner, & Schupp, 2017, p. 164). STI risk will subsequently be defined as the probability and the severity of the consequences of contracting an STI as a result of having sexual intercourse without a condom before both sexual partners are tested on STIs. Consequently, STI risk communication is expressing the probability and severity of the consequences of contracting an STI. The objective of STI risk communication is subsequently to make individuals
perceive the STI risk as likely to happen to them, and the consequences of it happening must be perceived as threatening (Wolfers et al., 2011).
Perceiving risk is as such a cognitive judgement based on an individual’s own beliefs about the likelihood of something happening to them, and the severity of the consequences (Lipkus, 2007; Veland & Aven, 2013). To explain the cognitive process of perceiving STI
risk, the Expectancy-Value Theory (EVT) is used. EVT describes the processes of motivating individuals towards specific behaviour based on the consideration of the expected
consequences of a particular behaviour (Wigfield & Eccles, 2000). Thus, from this can be deduced that when judging the STI risk communicated to them, individuals will weigh two aspects: the expectations that engaging in unprotected sexual intercourse before both partners are tested on STIs can lead to an STI and consider the seriousness of the consequences of infection with an STI.
EVT stresses the importance of both risk constructs being presented in a risk message, as only both will increase risk perceptions (Betsch, Haase, Renkewitz, & Schmid, 2015; Sheeran et al., 2014; Weinstein, 2000; Wigfield & Eccles, 2000). This suggests straight effects of the exposure to probability on the perceived susceptibility and value of the consequences on the perceived severity. As we are interested in the separate effects, the process of perceiving risk will be split upon its dual structure. In the following, these processes will be further explained.
STI Probability and Perceived STI Susceptibility
Probability in itself is an abstract term defined by Lipkus (2007) as the ‘likelihood of an event happening’ (p. 696). In the case of STI probability, it is the likelihood that you contract an STI when you have sexual intercourse without a condom before both sexual partners are tested for STIs. Important to note is that probability expresses a level of
uncertainty (Veland & Aven, 2012). In an example using percentages, one could illustrate that the level of certainty is always between 0% and 100%: never 100% or 0% because then it is no longer likely but rather a certainty that the presented risk will occur (Lipkus, 2007).
Uncertainty can be manipulated in its value (i.e. high or low values of probability), but as well in the format in which the values are presented (Lipkus, 2007; Visschers, Meertens, Passchier, & De Vries, 2009). The latter will be discussed first.
Probability format can be a numerical (e.g. ‘1 out of 5 chance on STI infection’), verbal (e.g. ‘very likely to get an STI’), or visual representations (e.g. using graphical
presentations of chance) (Lipkus, 2007; Mevissen et al., 2009; Wogalter, Young, Brelsford, & Barlow, 1999). The discussion of which format is the most outstanding to increase
susceptibility is subject to many literature review studies, yet no consensus is found (Visschers et al. 2009).
It can be argued that numerical probability information is preferred over verbal probability information when people need to evaluate risk (Visschers et al., 2009). Additionally, verbal probabilities have the disadvantage of being open to multiple interpretations (Fischhoff et al., 1993), while numerical probabilities are more precise representations of the chance of the risk occurring and hold a specific scientific credibility (Lipkus, 2007, p. 699).
However, one of the disadvantages of using numerical probability is the level of numeracy required of the participants (Lipkus, 2007; Visschers et al., 2009). Yet, level of numeracy, meaning the minimum required skills needed to understand and process numerical information, can be expected from the sample used in this study as it consists of students in higher education. They are exposed to numerical information on a daily basis and are therefore expected to have basic numeracy skills and to be able to process the numerical probability information (Visschers et al., 2009). Thus, the format of numerical probability information is most suitable for this study, as the goal is to increase perceived STI
susceptibility as a result of people evaluating the STI risk information presented to them. Furthermore, within a format of probability information, there are various possibilities. Combining multiple probability-formats eliminates the chances of wrongful processing due to biasing effects (Slovic, Monahan, & MacGregor, 2000).
Therefore, this study will apply a combination of odds and percentage probability information, as both prove to be effective in increasing risk perceptions (Slovic, Monahan, & MacGregor, 2000).
Based on the above, STI probability will be manipulated in high and low numerical probabilities expressing the likelihood of contamination with an STI. Returning to the EVT (Wigfield & Eccles 2000), one can argue that the judgement of a high STI probability will have a larger positive effect on perceived STI susceptibility of students. As the certainty that one can contract an STI after not using a condom is larger and therefore will increase
perceived susceptibility to the STI risk. The following hypothesis will test this assumption: H1: An STI risk message containing high STI probability information has a larger positive effect on perceived STI susceptibility of students, compared to low STI probability information.
A last note must be made regarding the measurement of perceived STI susceptibility. Students appear to be bad in correctly estimating one’s risk (Wolfers et al., 2011). Even more type of measurement assessment used can bias the estimate of perceived susceptibility
(Haase, Renkewitz, & Betsch, 2013). Studies mainly use one type of self-constructed measure: verbal Likert-scales (e.g., Carey & Sarma, 2016; Donné, Hoeks, & Jansen, 2017; Mevissen et al., 2011). It is therefore argued that there is a need for more elaborative measures of perceived susceptibility (Witte et al., 1998).
Windschitl and Wells (1996) argue that verbal measures are more fitting and sensitive to measure ‘psychological uncertainty’, as compared to numerical measures. Psychological uncertainty can be translated towards the perceived susceptibility, as it is in essences, a belief about one owns likelihood that engaging in risky behaviour will lead to the presented risk. Yet, a measuring susceptibility using verbal measurements does hold a certain level of ambiguity (Fischhoff et al., 1993). Haase et al., (2013) argue that the use of a visual analogue
scale can provide a more sensitive measure of change in perceived STI susceptibility. Therefore, this study will use two types of measurements to assess perceived STI
susceptibility: a verbal seven-point Likert scale (L) and a visual analogue scale (V). This will contribute to a more proper analysis of the effect STI risk communication on perceived STI susceptibility, and rule out any biases relating to scale measurement.
Severe STI Consequences and Perceived STI Severity
The consequences of contracting STIs are quite diverse. They range from the short-term bodily consequences of the STIs, which can be experienced; as painful (e.g. pain during urinating), to the long-term consequences affecting life-quality (e.g. infertility) (Soa Aids Nederland, 2017a). Whether the consequences of the risk are being perceived as severe is subjective and based on the judgement of personal relevance and importance (Fischhoff et al., 1993; Frank, Murphy, Chatterjee, Moran, & Baezconde-Garbanati, 2015; Wigfield &
Cambria, 2010). Relating perceived severity to the EVT, it poses that perceiving
consequences is affected by the expected value that a specific event holds, in whichvalue can be defined as containing personal relevance (Wigfield & Cambria, 2010). If the presented consequences of a risk are not important to the individual, the consequences will not have any value, and therefore they will not be perceived as severe.
This stresses the importance of manipulating the STI consequences based on personal relevance. Personal relevance can be established by selecting those consequences that are important to the target group of the health message, as opposed to general consequences (Frank et al., 2015). In this study, the STI consequences need to be perceived as severe by students in higher education. Hence, a pre-test was conducted to select those STI
consequences that are perceived as highly severe and as less severe by the target group (results of the pre-test are presented in Appendix D). Thus it can be assumed that students who receive STI consequences that are judged by them as personally highly severe will have
a more positive effect as compared to presenting them with severe STI consequences that are being judged as less personally relevant (low perceived severity). The following hypothesis tests this assumption:
H2: An STI risk message containing high severe STI consequences has a larger positive effect on perceived STI severity of students, as compared to low severe STI consequences.
Cross Manipulation and Interaction Effects
Next to the main effects proposed above, another angle should be examined. One point of criticism on previous studies examining risk communication is the insufficient focus on how the underlying processes interplay. According to EVT, both probability and severe consequences need to be present to increase risk perceptions (Sheeran et al., 2014; Wigfield & Eccles 2000). Yet, when not specifying on those ‘risk perceptions’ the process appears to be straight-lined; probability information will make individuals perceive themselves as more susceptible and severe STI consequences will make them perceive the consequences as severe (Weinstein, 2000). Although these constructs are closely related it is limited to assume that the effects of the separate manipulated risk constructs are straight-lined (Harris, Corner, & Hahn, 2009). Even more, a health message normally contains at least both risk constructs, thus the question arises whether these interact on the separate risk constructs or function individually next to each other on separate risk perceptions (Aven, 2011; Sheeran et al., 2014).
To our knowledge, no previous research examined any cross manipulation or interaction effects on the separate risk perceptions. As in an endeavour to map how the manipulation of STI probability and severe STI consequences individually and in interaction affects the separate risk perceptions the following three sub-research questions will be proposed:
RQa: How does the manipulation of STI probability information in an STI risk message affect perceived STI severity?
RQb: How does the manipulation of severe STI consequences in an STI risk message affect perceived STI susceptibility?
RQc: How does the interaction between the manipulated STI probability and severe STI consequences affect perceived STI severity and perceived STI susceptibility?
Message Evaluation and Message Induced Emotions
To ensure that the results of the study reveal the precise effect of the manipulation STI risk on the separate risk perceptions, two control elements were included based on the
following arguments. First, the manipulation of STI risk is presented in the format of a health message and studies show that characteristics of the message can affect how a message is perceived and thus affect the related risk perceptions (Mevissen et al., 2009; Ferrer & Klein, 2015; Lipkus, 2007). Thus to control whether these message characteristics were not
evaluated differently for the manipulation of STI risk, the message will be evaluated in its credibility, personal relevance, novelty, comprehensibility, exaggeration, and readability.
Second, the focus of this study lies in the perception of risk, thus relates to cognitive processing. And although some argue that perceiving risk can be mainly a cognitive
assessment (Freimuth, Jamison, Hancock, Musa, Hilyard, & Quinn, 2017), there is research suggesting that perceiving risk is strongly related to affective processing (Ferrer & Klein, 2015; Fischhoff et al., 1993; Schmälzle et al., 2017; Wolfers et al., 2011). This would mean that manipulating STI risk would as well affect emotions. Yet, the focus of this study is a cognitive aspect. An additional check will control how the manipulation of STI risk affects an individual’s emotions. Six emotions will be monitored; fear, sadness, worry, hope,
Methodology Experimental Design
For this online survey-embedded experiment a 2 (STI probability: high vs. low) by 2 (severe STI consequences: high vs. low) between-subjects factorial design is used to examine the effects of STI risk manipulation on the perceived STI susceptibility and perceived STI severity of students in higher education (see Table 1 for the factorial design, Appendix A).
Sample
A total of 288 participants entered the online experiment. Participants who did not agree with the informed consent (n = 2), and those who withdrew their consent after the debriefing were excluded (n = 8). Subsequently, participants who did not complete the survey until the STI risk perception measures (n = 49) were excluded. To qualify to participate in the study, participants had to be sexually active, as sexual experience can influence personal risk perceptions (Wolfers et al., 2011).Hence, participants who indicated to be not sexually active in the past six months were excluded (n = 31). Furthermore, participants who reported to be homosexually orientated (female-female) were excluded (n = 3), as they are usually not in a position to use condoms when having sexual intercourse and therefore need specifically tailored STI risk communication (Diamant, Lever, & Schuster, 2000). The sample should represent students in higher education; thus non-students were excluded (n = 4). At last, participants who took less than 10 seconds to read the presented stimulus were excluded (n = 33), based on the assumption that they could not have attentively perceived the presented information.
The final sample of the study consists of 158 participants of which 67.7% women (n_women = 107 and n_men = 51). The average age of the sample is 21.67 years old (SD = 2.41), with a minimum age of 18 and maximum age of 33. The majority of the participants are currently following a Master of Science (40.5%), followed by a Professional Bachelor
(31.0%), Bachelor of Science (19.6%), Bachelor of Arts (5.1%), Post Graduate (2.5%), and Master of Arts (1.3%). Most of the participants in the study hold the Belgian nationality (n = 123), followed by Dutch (n = 34) and German (n = 1) nationality. Heterosexual orientation was the dominant sexual orientation in the sample (91.8%), followed by homosexual (male-male) (4.4%) and bisexual (3.2%). Most of the participants had sexual intercourse with a steady partner (n = 101) and 56 with one or more casual partner(s). Most of the participants indicated to never use a condom (45,6%) before both sexual partners were tested for STIs. Furthermore, 17.7% of the participants indicated to have been infected with at least one STI in the past, of which chlamydia was the most common STI (n = 14), followed by HPV (n = 5), genital warts (n = 3), herpes (n = 3), gonorrhoea (n = 14), and scabies (n = 1). 55.7% of
the participants have never been infected with an STI, and 25.9% of participants did not know, as they were never tested. All questions had the answer category ‘I would rather not answer'. Participants who indicated the latter were indicated as missing values. An overview of sample characteristics is presented in Table 2 (see Appendix B).
Procedure
Data was collected through a convenient sampling among students in higher education in the Netherlands and Belgium, between the 22nd of November 2017 and 7th of December 2017. Participants were made aware of the study through announcements on Facebook informing them about the topic and duration of the study. All participants entered the online experiment through an anonymous link and did not receive any compensation for their participation.
Once participants entered the online experiment, they were briefly introduced to the topic of the study. After the informed consent procedure, participants received basic questions about their demographical and sexual behavioural characteristics. Subsequently, participants were asked to read the thereafter-presented STI risk message attentively. The participants
were randomly assigned to one of four manipulated STI risk messages. After the manipulation, participants answered questions about their STI risk perceptions, the
manipulation check, and two additional questions evaluating the message characteristics and the induced emotions. At the end of the survey, participants were debriefed about the
manipulated STI risk information and given the option to opt-out of the study. The full debriefing is included in Appendix C.
Materials
Pre-test. A pre-test with 48 participants was conducted to examine the perceived
severity of the different consequences of contracting an STI. 46 items, representing the negative bodily consequences (e.g. penis cancer) and emotional consequences of STIs (e.g. shame), were assessed on perceived severity. A more elaborate description of the set-up and selection of the STI consequences is included in Appendix D, as well as an overview of all items used (see Table 3).
Mean-scores and frequencies were examined for the gender-neutral consequences (see Table 4a, Appendix D), and subsequently for women and men separately (see Table 4b and Table 4c, Appendix D). Paralysation, meningitis, anus, throat or liver cancer, and cancer to the reproductive organs were selected for the high severe STI conditions and throat infection, fatigue, headache, and lower self-esteem were selected for the low severe conditions.
Stimulus Materials. The STI risk message consists of general STI information, STI
probability information and some of the consequences of getting an STI. The general STI information was similar in all four conditions and explained what an STI is, how it is transferred and mentioned the most common STIs in the Netherlands. STI probability
information was manipulated by adding high or low probability information. For example low probability of contracting an STI was expressed by telling the following: ‘... statistics show that not consistent using condoms delivers a chance of around 22% to contract an STI in 1 out
of 5 students’, as compared to high probability information: ‘... statistics show that not consistent using condoms delivers a chance of around 82% to contract an STI in 4 out of 5 students'. Additionally, the consequences of getting an STI were manipulated based on the results of the pre-test by presenting either STI consequences that are perceived as low in severity (e.g. a headache) or STI consequences that are perceived as high in severity (e.g. meningitis). This leads to the following four conditions: X1Y1 (low STI probability x low severe STI consequences), X2Y1 (low STI probability x high severe STI consequences), X1Y2 (high STI probability x low severe STI consequences), and X2Y2 (high STI probability x high severe STI consequences). The stimulus material is included in Appendix E.
Measurements
Background Characteristics. Demographic variables measured were age,
nationality, gender, education, and sexual orientation. Additional variables asked participants about their sexual activity, sexual partner type, condom use, and STI history (and if
applicable, type of STI). Specific measurements and categories of the background variables are included in Appendix F.
Perceived STI Susceptibility. Two measurement tools assessed perceived STI
susceptibility. First, perceived STI susceptibility (L) is measured by two items rated on a seven-point Likert scale from 1 = ‘completely disagree’ to 7 = ‘completely agree’. For example ‘If I would have sex without a condom before me and my sexual partner are both tested for STIs then it is very likely that I get infected with an STI’. This dependent variable is computed by the mean of the summation of the values of the separate items and showed acceptable reliability (Cronbach’s Alpha = .70, M = 4.24, SD = 1.31). All items and scale constructions are included in Appendix F.
The second perceived susceptibility measure, perceived STI susceptibility (V), is assessed with the use of a visual analogue slider. Participants were asked the following
question: ‘If you would have sex without a condom, before you and your sexual partner are tested on STIs, how big do you estimate your chance of getting infected with an STI?’
Answers were indicated on a visual analogue slider going from 0 = ‘no chance at all’ to 100 = ‘definitely chance on infection’.
Perceived STI Severity. Perceived STI severity is measured with four items based on
the scale of de Bruijn, Spaans, Jansen, and van’t Riet (2016). The items were measured on a seven-point Likert scale from 1 = ‘completely disagree’ to 7 = ‘completely agree’. For example: ‘The negative consequences of sex without a condom, such as infection with an STI, are for me serious’ (see Appendix F for all items and scale construction). Perceived STI severity, this dependent variable is as well computed by the mean of the summation of the values of the separate items and showed acceptable reliability with a Cronbach’s Alpha of .77 (M = 5.61, SD = 1.08).
Message Evaluation. Message evaluation was assessed by six variables judged on a
seven-point Likert scale from 1 = ‘completely disagree’ to 7 = ‘completely agree’. The
message evaluation variables were credibility, personal relevance, novelty, comprehensibility, exaggeration, and readability. The following statement preceded the judgement of these evaluations: ‘The message about the chance and severity of STI infection was...’ (1 =
‘credible’; 2 = ‘personally relevant to me’; 3 = ‘unknown information; 4 = ‘understandable’; 5 = ‘exaggerated’; 6 = ‘readable’). The variables for the message evaluation were based on the studies of Mevissen et al. (2009) and (Lipkus, 2007).
Message Induced Emotions. Message induced emotions were assessed by six
emotions judged on a seven-point Likert scale from 1 = ‘completely disagree’ to 7 = ‘completely agree’. The following statement preceded the judgement of these emotions: ‘During or shortly after reading the message I personally felt...’. (1 = ‘scared’; 2 = ‘sad’; 3 = ‘worried’; 4 = ‘hopeful’; 5 = ‘satisfied’; 6 = ‘happy’).
Results Randomization Checks
One-way Multivariate Analysis of Variance (MANOVA) and Pearson Chi-square tests were conducted to assess whether participants’ demographical (i.e. age, nationality, gender, and education) and sexual behavioural characteristics (i.e. condom use, sexual partner type, and STI history) were equally distributed between the different experimental conditions. The results showed, except for education, no significant differences in distribution across the conditions (all p > .224). The results of the separate randomisation checks are included in Appendix G. Education was not equally distributed between the high and low severe STI consequences conditions (𝛸2 = 15.43, p = .009), therefore the variable will be included as a covariate in the main analyses.
Manipulation Checks
In order to assess whether participants perceived the manipulation as intended two manipulation checks were conducted. Participants were exposed to two statements and asked to rate them on a seven-point Likert scale from 1 = ‘completely disagree’ to 7 = ‘completely agree’. The first statement assessed the manipulation of STI probability, i.e. ‘The
consequences of having sexual intercourse without a condom, before both sexual partners are tested on STIs, can happen to me'. The second statement assessed the manipulation of severe STI consequences, i.e. ‘The consequences of having sexual intercourse without a condom, before me and my sexual partner are both tested on STIs, are severe’.
The manipulation check was conducted by a two-way MANOVA, with STI probability and severe STI consequences as factors and the two manipulation checks of susceptibility and severity as dependent variables. The results indicated a failed manipulation for STI probability on the manipulation check of susceptibility (F(1,141) = .95, p = .333, η2 =
.01), and for the severe STI consequences on the manipulation check of severity (F(1,141) = .15, p = .701, η2 = .00). Comparisons of the mean-scores are presented in Table 5 (see
Appendix H).
Statistical Analyses
For all statistical analyses, IBM SPSS 23.0 was used. Examination of the dependent variables showed no significant correlation between perceived STI susceptibility (V) and perceived STI severity (r = .06, p = .432). Therefore, the analyses were split into a two-way MANOVA and a two-way Analysis of Variance (ANOVA) to examine the effects of the STI risk manipulation on the STI risk perceptions. The two-way MANOVA was conducted with STI probability and severe STI consequences as factors and perceived STI susceptibility (L) and perceived STI severity as dependent variables (r = .20, p = .011). The two-way ANOVA was conducted with STI probability and severe STI consequences as factors and perceived STI susceptibility (V) as the dependent variable. All analyses were tested with a significance level of p < .050.
Before the main analyses were conducted, two pre-analyses were conducted to
examine the effect of the manipulation of STI probability and severe STI consequences on the message evaluation and message induced emotions. The variables that were significantly affected by the manipulated STI risk were subsequently added as covariates in the follow-up analysis to control whether the initially found effects remained.
Pre-Analyses
To test the effect of the manipulated STI probability and severe STI consequences on the message evaluation (i.e. credibility, personal relevance, novelty, comprehensibility, exaggeration, and readability) and message induced emotions (fear, worry, sadness,
satisfaction, hope and happiness) two times a six six-way MANOVA was conducted with STI probability and severe STI consequences as factors, and the message evaluations and message
induced emotions as dependent variables. The results of the pre-analyses are presented in Appendix I.
In sum, based on the message evaluations, all manipulated messages were on average perceived as personally relevant, highly credible, comprehensible, and readable andwere not perceived as exaggerated or holding novel information (an overview of mean-scores is
presented in Table 7, Appendix I). Of the message induced emotions; sadness, happiness were low and hope, fear and satisfaction showed a low moderate level over all conditions (an overview of mean-scores is presented in Table 9, Appendix I). Yet, readability, personal relevance, novelty, fear, and satisfaction, were affected by the manipulation of at least one of the STI risk constructs or by the interaction term (all p < .044) and therefore included as covariates in the follow-up analysis.
Main Analyses
The demographical variable education was added as a covariate in the based on the results of the randomization check. Both analyses showed no significant interaction effect of education on one of the dependent variables (all p > .217) (see Table 10, Table 11, and Table 12 in Appendix J, for the multivariate and univariate test results).
Manipulation of STI probability. Results of the two-way MANOVA revealed a
small significant multivariate effect of STI probability on the combined dependent variables, F(2, 151) = 3.56, p = .031, Wilk’s Λ = .96, η2 = .05. The subsequent univariate ANOVAs
tested the first hypothesis and the proposed cross manipulation effect of STI probability on perceived STI severity.
In accordance with the first hypothesis, the results revealed a small significant main effect of STI probability on perceived STI susceptibility (L), F(1,152) = 4.34, p = .039 η2 =
.03. Students who received the high STI probability information perceived themselves more susceptible for contracting an STI, as compared to those who received the low STI probability
information. These results are as well confirmed by the additional two-way ANOVA, which revealed a large significant main effect of STI probability on perceived STI susceptibility (V), F(1,152) = 31.54, p < .000, η2 = .17. The mean-score comparisons are presented in Table 13
(see Appendix J). Note that the results of the second measurement should be interpreted with caution, as the check for homogeneity of variance was not assumed based on the Levene’s test, F(3,153) = 8.25, p < .000. Hence based on the results, the first hypothesis is confirmed.
Subsequently, the proposed cross manipulation effect of STI probability on perceived STI severity was tested. Results showed no significant effect of STI probability on perceived STI severity, F(1,152) = 1.30, p = .256, η2 = .01. Hence, no cross effect was found for the
manipulation of STI probability on perceived STI severity.
Manipulation of severe STI consequences. Results of the two-way MANOVA
showed a marginally significant multivariate effect found of the severe STI on the combined dependent variables, F(2, 151) = 2.46, p = .089, Wilk’s Λ = .97, η2 = .03. The subsequent
univariate ANOVAs tested the second hypothesis and the proposed cross manipulation effect of severe STI consequences on perceived STI susceptibility.
The second hypothesis examined the main effect of the manipulation of severe STI consequences on perceived STI severity. The univariate ANOVA showed no significant main effect, F(1,152) = 1.49, p = .224, η2 = .01. Hence, the second hypothesis is not confirmed.
Subsequently, the proposed cross manipulation effect of severe STI consequences on perceived STI susceptibility was tested. No significant effect was found on perceived STI susceptibility (L) (F(1,152) = 2.30, p = .131, η2 = .02), nor on the second measure of
perceived STI susceptibility (V) (F(1,152) = 1.97, p = .162, η2 = .01). Hence, no cross effect
was found for the manipulation of severe STI consequences on perceived STI susceptibility.
Interaction Effects. Results of the two-way MANOVA and the two-way ANOVA
the combined dependent variables, F(2, 151) = 1.72, p = .182, Wilk’s Λ = .98, η2 = .02. The
subsequent univariate ANOVAs showed also no significant interaction effects on perceived STI susceptibility (L) (F(1,152) = .00, p = .979, η2 = .00), perceived STI susceptibility (V)
(F(1,152) = .02, p = .898, η2 = .00), or on perceived STI severity (F(1,152) =3.74, p = .071, η2
= .02).
Main Analyses with Covariates
The follow-up two-way MANOVA and two-way ANOVA were conducted with covariates readability, personal relevance, novelty, fear, and satisfaction. The aim was to control whether the effects of the initial analyses remained. Two notes must be made. First, only 145 participants of the initial sample (N = 158) completed the assessment of message evaluation and message induced emotions. Second, the check for homogeneity of variance was not assumed in the MANOVA for perceived STI susceptibility (V) (F(3,141) = 5.21, p = .002), and perceived STI severity (F(3,141) = 4.09, p = .008).
Manipulation of STI probability with Covariates. The multivariate tests showed no
significant effect of STI probability on the combined dependent variables, F(2,134) = 2.18, p = .117, Wilk’s Λ = .97, η2 = .03, after controlling for the covariates. The univariate ANOVAs
are discussed to control whether the results found in the initial analyses remained.
Contradicting the first hypothesis, results showed no significant main effect of STI probability on perceived STI susceptibility (L), F(1,135) = 1.86, p = .175, η2 = .01. However,
this finding was refute by the second measurement of perceived STI susceptibility, which showed that STI probability did have a large significant effect on perceived STI susceptibility (L), F(1,135) = 25.89, p = .000, η2 = .16. Students that received high STI probability
information perceived themselves more susceptible for contracting an STI, as compared to those who received the low STI probability information. The mean-score comparisons are
presented in Table 14 (see Appendix K). Hence, the first hypothesis can only be partially supported when controlled for the covariates.
The proposed cross manipulation effect of STI probability on perceived STI severity remained insignificant after controlling for the covariates, F(1,135) = 1.51, p = .221, η2 = .01.
Manipulation of severe STI consequences with Covariates. The multivariate tests
showed a small marginal significant effect of severe STI consequences on the combined dependent variables (F(2,134) = 2.49, p = .087, Wilk’s Λ = .96, η2 = .04).
The subsequent univariate ANOVA showed no significant main effect of severe STI consequences on perceived STI severity was found after controlling for the covariates, F(1,135) = .48, p = .489, η2 = .00. Hence, the second hypothesis remains unsupported after
controlling for the covariates.
The proposed cross manipulation effect of severe STI consequences on perceived STI susceptibility was partially confirmed. A small marginal significant cross manipulation effect of severe STI consequences on perceived STI susceptibility (L), F(1,135) = 3.64, p = .059, η2
= .03. Students receiving the low severe STI consequences perceived themselves more susceptible for contracting an STI, as compared to those receiving high severe STI consequences. The mean-score comparisons are presented in Table 14 (see Appendix K). However, this was not confirmed by the second measurement of perceived STI susceptibility (L), F(1,135) = 1.68, p = .198, η2 = .01.
Interaction Effects with Covariates. The multivariate test showed no significant
difference of the interaction term on the combined dependent variables, F(2,134) = 1.39, p = .252, Wilk’s Λ = .98, η2 = .02. This was confirmed by the following univariate ANOVAs. The
interaction between STI probability and severe STI consequences showed no significant effects on perceived STI severity (F(1,135) = 2.45, p = .120, η2 = .02), on perceived STI
susceptibility (L) (F(1,135) = .04, p = .834, η2 = .00), or on perceived STI susceptibility (V)
(F(1,134) = .09, p = .769, η2 = .00).
Worry. The multivariate test showed medium significant effect of worry on the
combined dependent variables, F(2,134) = 4.04, p = .020, Wilk’s Λ = .94, η2 = .06. The
subsequent univariate ANOVAs indicated that worry had a significant effect on perceived STI severity (F(1,135) = 5.70, p = .018, η2 = .04), yet not on perceived susceptibility (F(1,134) =
.93, p = .336, η2 = .01).
Discussion and Conclusion
The goal of the study was to examine how the manipulated STI probability and severe STI consequences individually and in interaction affected perceived STI susceptibility and perceived STI severity of students in higher education. The results of the pre-analyses indicate that the manipulation of STI risk not only affects the risk perceptions but as well message evaluation and message induced emotions. Manipulation of the risk constructs affected for one how a message is evaluated in readability, personal relevance and novelty, and secondly revealed different effects in the emotions fear, worry, and satisfaction. These variables were included as covariates in a follow-up analysis of the main effects to ensure that it is the manipulations of STI risk which induces the effects found on the risk perceptions.
The first hypothesis tested whether an STI risk message containinghigh STI
probability information made students perceive themselves as more susceptible to contracting an STI as compared to low STI probability information. The first hypothesis was confirmed by both measurements of perceived STI susceptibility in the initial analysis. However, after controlling for the covariates, the effect partially disappeared. Only the visual analogue measurement of perceived susceptibility proved to be accurate enough to indicate a significant difference of high and low STI probability on perceived susceptibility. These findings emphasise two important things. First, the finding is in line with the assumption
made based on the EVT (Weinstein, 2000, Wigfield & Eccles 2000), which poses that high probability information increases the expectation that the risk will happen and therefore has a larger effect than low probability information. Second, this finding supports Haase et al.’s (2013) line of reasoning that a visual analogue scale might be more precise in measuring perceived susceptibility when exposed to numerical probability information. When the covariates were included, the latter indicated a more precise measure of perceived susceptibility.
The second hypothesis proposed that communicating high severe STI consequences compared to low severe STI consequences would make students perceive the consequences of contracting an STI as more severe. Results indicated no significant differences in perceived STI severity for the high and low severe STI consequences in the initial analysis and after controlling for the covariates. The lack of significant results could be due to the failed manipulation of the high and low severe STI consequences. Although an elaborate pre-test was conducted to divide the STI consequences into two conditions: high and low severe (see Appendix D), the result of the experiment indicated that all STI consequences on average were perceived as highly severe. This finding reveals that the difference between the high and low condition was not large enough to distinctively increase perceived STI severity.
Additionally the pre-analysis indicated that both conditions do not significantly differ in personal relevance. The STI consequences were pre-tested in a similar population as the study sample, as their judgement of the items would be specific to the target group and therefore more relevant to them as opposed to consequences perceived as severe by the general population (Frank et al., 2015). Therefore it was assumed that the high severe consequences would be more personally relevant as opposed to low severe consequences (Wigfield & Cambria, 2010). Nevertheless, no significant differences were found between the two conditions for personal relevance.
The proposed cross manipulation effects on the separate risk perceptions could only be partially confirmed. The initial analysis and follow up analysis revealed students perceived both the low and high probabilities of contracting an STI as highly severe. Furthermore, the initial analysis did not reveal a cross manipulation effect of the severe STI consequences on perceived STI susceptibility. The follow-up analysis refutes this by revealing a marginal effect indicating that communicating high severe STI consequences will make students perceive themselves as less susceptible to contracting an STI, as when they were exposed to low severe STI consequences. The process of optimistic bias can be supportive in explaining this effect person’ (Harris & Hahn, 2011; Schmälzle et al., 2017; Weinstein, 1982).
Optimistic bias (i.e. unrealistic optimism) entails the following: ‘people rate negative future events as less likely to happen to themselves than to the average person and positive events as more likely to happen to themselves than to the average person’ (Harris & Hahn, 2011, p.135). This could explain that when students are exposed to high severe STI consequences, the process of optimistic bias can lead to lower feelings of susceptibility, as they believe that high severe consequences are less likely to happen to them as to others. This argument should be made with caution. The process of unrealistic optimism in itself not always supported, as it can be argued that some individuals will correctly estimate their level of susceptibility due to not engaging in the risky behaviour (Harris, Corner, & Hahn, 2009; Harris & Hahn, 2011). This would indicate that their optimism is not subject to bias (Fishhoff et al., 1993; Harris & Hahn, 2011). Therefore the process of optimistic bias and its role in the cross manipulation effect of severe STI consequences on perceived susceptibility is a crucial topic for future research.
A second remark must be made as the finding was only supported by the Likert scale measuring perceived susceptibility. Windschitl and Wells (1996) argue that this type of measurement is better when judging a ‘psychological uncertainty’. Relating to the optimistic
bias about the negative events, one might assume that measuring perceived susceptibility to a negative event is a bigger psychological uncertainty than the measuring perceived
susceptibility as an assessment of a numerical probability impulse. Yet, to our knowledge the comparison between scales is not yet researched, on that account indicates a vital subject for further examination.
At last, both analyses revealed no significant interaction effects. The manipulation of probability and severe STI consequences in high or low-value combinations elicited on average the same level of perceived susceptibility and perceived severity. This might indicate that interaction between the manipulated constructs is not necessary to increase the separate STI risk perceptions. Note that this does not mean that the risk constructs should not appear in the same health message, as according to EVT they are both necessary to increase ‘risk
perceptions’ (Weinstein, 2000). Investigating the interaction effects on the individual risk perceptions is still in an exploratory phase and needs further examination.
Limitations
A general remark regarding generalizability must be made. Examining risk and the related risk perceptions in the field of health communication are specific to the health problem they are representing (Ferrer & Klein, 2015; Fischhoff et al., 1993; Freimuth et al., 2017). Therefore these results are specific for the communication of STI risk, and its effects on the risk perceptions of a student population in the Netherlands and Belgium. Note that the sample was only a small fraction of the whole student population therefore further
examination in a large sample is strongly encouraged.
Two additional limitations of the study must be mentioned. First, the manipulation checks to indicate differences in the manipulated risk constructs failed. As the results of the main analyses showed that measuring perceived susceptibility is sensitive to measurement tool, the first reason for the failed manipulation check for perceived susceptibility can be the
due to the limited measurement in the manipulation check assessment. This urges other studies to consider and pre-test their measurement tools as they can generate a level of bias (Betsch, Haase, Renkewitz, & Schmid, 2015; Fishhoff et al., 1993). As for the failed manipulation check of severe STI consequences, the same argument can be made as for the failed main effect (cf. supra): the selected STI consequences did not differentiate enough in perceived severity, and therefore there was no significant difference between the high and low severe STI condition.
Second, this study did not include a control group in the experimental design. The decision to exclude a control group was based taking into consideration the requested time of participating students. Including a control group would require an additional string of
variables measuring both control and STI risk perceptions, resulting in a longer experimental study. As this was an online experiment limiting time can be of essence to ensure the focus of participating students. This does not eliminate that it limits the interpretation of the findings to a mere comparison of high against low values of the STI risk constructs. Although many studies show that exposing students to risk information is effective in increasing risk
perceptions (Mevissen et al., 2009; Sheeran et al., 2014), these studies do not specify on the manipulation of STI risk value (high versus low) on the separate risk perceptions. Future studies should include a control group to have a neutral basis for comparison.
Implications and Future Research Directions
The first implication made is the importance of emotions when evaluating risk. Although risk was constructed to trigger cognitive processing, results revealed that being exposed to risk also influences affective processing. When readability, personal relevance, novelty, fear, worry and satisfaction were added as covariates the initial findings were subject to change, indicating that the relationship between communicating STI risk and the risk perceptions is sensitive to the message evaluations and message induced emotions (Mevissen
et al., 2009; Ferrer & Klein, 2015; Lipkus, 2007). The coherence between cognitive and affective processing is in line with arguments made by researchers in the field of risk communication (e.g.: Ferrer & Klein, 2015; Fischhoff et al., 1993; Schmälzle et al., 2017; Visschers, Wiedemann, Gutscher, Kurzenhäuser, Seidl, Jardine, & Timmermans, 2012; Wolfers et al., 2011). Especially worry appeared to be a significant covariate on the perceived STI severity. Worry has been studied in previous research as a related factor to the risk
perceptions (Wolfers et al., 2011). Yet, to our knowledge, its influence in the relationship between the manipulation of the STI risk constructs and the separate STI risk perceptions has not been studied in-depth. Therefore future research should further examine the role of worry, and emotions in general, related to perceiving STI risk.
The above indicates a second implication: the importance of splitting the risk
perceptions in perceived severity and perceived susceptibility. As although perceived severity and perceived susceptibility are moderately related (r = .20), perceiving them as one variable might lead to misinterpretation of results and ineffective communication of risk.
Additionally, two practical implications are addressed. First, manipulating severe STI consequences in low and high values appeared to be not so self-evident. However, it is crucial, as presenting students with high severe STI consequences will make them perceive themselves as less susceptible towards contracting an STI when having unprotected sex. In this study, we focused on the actual bodily and emotional consequences that arise when contaminated with an STI. Though, STI consequences can as well be categorised in sexually transmitted disease (STD), e.g., chlamydia or HIV. An interesting point of view in different terms of value, as both diseases are sexually transmittable but have a different value in
severity (Barth et al., 2002; Newby et al., 2012). Thus, future research should further examine how these different STDs can influence the effect on perceived susceptibility, and although not confirmed in this study, the effect on perceived severity.
Second, next to the difference in severity of STIs, not all STIs are as likely to occur. This entails that STIs hold both a level of severity and probability of occurrence. The decision to, for example, present STI risk information containing chlamydia, which has a high
probability to occur, but is less severe as for example HIV (Mevissen et al., 2009), could have a different effect on the separate risk perceptions. These types of interactions are to our knowledge understudies and need more attention.
To conclude, manipulating STI risk and its effect on the separate risk perceptions is still in an exploratory phase. These findings are a glimpse into how the manipulation of STI probability and severe STI consequences can affect the separate risk perceptions. At the same time, they stress the need for further research examining the separate constructs of risk and the effects on the separate risk perceptions. Perceiving risk is a complex process, and for STI risk it can be concluded that students do not only rely on cognitive processing but as well on affective processing.
References
Aven, T. (2011). On some recent definitions and analysis frameworks for risk, vulnerability, and resilience.Risk Analysis, 31(4), 515-522.
Aral, S. O. (2001). Sexually transmitted diseases: magnitude, determinants and consequences. International journal of STD & AIDS, 12(4), 211-215.
Barth, K. R., Cook, R. L., Downs, J. S., Switzer, G. E., & Fischhoff, B. (2002). Social Stigma and Negative Consequences: Factors That Influence College Students' Decisions to Seek Testing for Sexually Transmitted Infections. Journal of American College Health, 50(4), 153-159.
Betsch, C., Haase, N., Renkewitz, F., & Schmid, P. (2015). The narrative bias revisited: What drives the biasing influence of narrative information on risk perceptions? Judgment & Decision Making, 10(3), 241-264.
Carey, R. N., & Sarma, K. M. (2016). Threat appeals in health communication: messages that elicit fear and enhance perceived efficacy positively impact on young male drivers. BMC Public Health, 16, 645.
Cates Jr, W., & Steiner, M. J. (2002). Dual protection against unintended pregnancy and sexually transmitted infections: what is the best contraceptive approach? Sexually transmitted diseases, 29(3), 168-174.
Centers for Disease Control and Prevention (2016). Sexually Transmitted Disease
Surveillance 2016. Atlanta: U.S. Department of Health and Human Services; 2017. Retrieved from: https://www.cdc.gov/std/stats
Crosby, R., & Bounse, S. (2012). Condom effectiveness: where are we now? Sexual Health, 9(1), 10-17.
de Bruijn, G.-J., Spaans, P., Jansen, B., & van’t Riet, J. (2016). Testing the effects of a message framing intervention on intentions towards hearing loss prevention in adolescents. Health Education Research, 31(2), 161-170.
Deroche, T., Stephan, Y., Woodman, T., & Le Scanff, C. (2012). Psychological Mediators of the Sport Injury—Perceived Risk Relationship. Risk Analysis, 32(1), 113-121. Diamant, A. L., Lever, J., & Schuster, M. A. (2000). Lesbians' Sexual Activities and Efforts
to Reduce Risks for Sexually Transmitted Diseases. Journal of the Gay and Lesbian Medical Association, 4(2), 41-48.
Donné, L., Hoeks, J., & Jansen, C. (2017). Using a narrative to spark safer sex communication. Health Education Journal, 76(6), 635-647.
Downing-Matibag, T. M., & Geisinger, B. (2009). Hooking Up and Sexual Risk Taking Among College Students: A Health Belief Model Perspective. Qualitative Health Research, 19(9), 1196-1209.
Fischhoff, B., Bostrom, A., & Quadrel, M. J. (1993). Risk Perception and Communication. Annual Review of Public Health, 14(1), 183-203.
Floyd, D. L., Prentice-Dunn, S., & Rogers, R. W. (2000). A Meta-Analysis of Research on Protection Motivation Theory. Journal of Applied Social Psychology, 30(2), 407-429. Frank, L. B., Murphy, S. T., Chatterjee, J. S., Moran, M. B., & Baezconde-Garbanati, L.
(2015). Telling Stories, Saving Lives: Creating Narrative Health Messages. Health Communication, 30(2), 154-163.
Freimuth, V. S., Jamison, A., Hancock, G., Musa, D., Hilyard, K., & Quinn, S. C. (2017). The Role of Risk Perception in Flu Vaccine Behavior among African‐ American and White Adults in the United States.Risk Analysis,37(11), 2150-2163.
Haase, N., Renkewitz, F., & Betsch, C. (2013). The Measurement of Subjective Probability: Evaluating the Sensitivity and Accuracy of Various Scales. Risk Analysis, 33(10), 1812-1828.
Harris, A. J., Corner, A., & Hahn, U. (2009). Estimating the probability of negative events.Cognition,110(1), 51-64.
Harris, A. J., & Hahn, U. (2011). Unrealistic optimism about future life events: A cautionary note.Psychological review,118(1), 135.
Hood, J. E., Hogben, M., Chartier, M., Bolan, G., & Bauer, H. (2014). Dual contraceptive use among adolescents and young adults: correlates and implications for condom use and sexually transmitted infection outcomes. Journal of Family Planning and
Reproductive Health Care, 40(3), 200-207.
Lipkus, I. M. (2007). Numeric, Verbal, and Visual Formats of Conveying Health Risks: Suggested Best Practices and Future Recommendations. Medical Decision Making, 27(5), 696-713.
Mevissen, F. E. F., Meertens, R. M., Ruiter, R. A. C., Feenstra, H., & Schaalma, H. P. (2009). HIV/STI Risk Communication: The Effects of Scenario-based Risk Information and Frequency-based Risk Information on Perceived Susceptibility to Chlamydia and HIV. Journal of Health Psychology, 14(1), 78-87.
Mevissen, F. E. F., Ruiter, R. A. C., Meertens, R. M., & Schaalma, H. P. (2010). The effects of scenario-based risk information on perceptions of susceptibility to Chlamydia and HIV. Psychology & Health, 25(10), 1161-1174.
Mevissen, F. E. F., Ruiter, R. A. C., Meertens, R. M., Zimbile, F., & Schaalma, H. P. (2011). Justify your love: Testing an online STI-risk communication intervention designed to promote condom use and STI-testing. Psychology & Health, 26(2), 205-221.
Montanaro, E. A., & Bryan, A. D. (2014). Comparing Theory-Based Condom Interventions: Health Belief Model Versus Theory of Planned Behavior. Health Psychology, 33(10), 1251-1260.
Moscou, S. (2016). Meeting Contraceptive Needs of College Students: Bringing Evidence Into Practice. The Journal for Nurse Practitioners, 12(1), e11-e15.
Moyer-Gusé, E., Chung, A. H., & Jain, P. (2011). Identification With Characters and Discussion of Taboo Topics After Exposure to an Entertainment Narrative About Sexual Health. Journal of Communication, 61(3), 387-406.
Nan, X., Dahlstrom, M. F., Richards, A., & Rangarajan, S. (2015). Influence of Evidence Type and Narrative Type on HPV Risk Perception and Intention to Obtain the HPV Vaccine. Health Communication, 30(3), 301-308.
Newby, K. V., Wallace, L. M., & French, D. P. (2012). How do young adults perceive the risk of chlamydia infection? A qualitative study. British Journal of Health Psychology, 17(1), 144-154.
Newton, D. C., & McCabe, M. P. (2005). The impact of stigma on couples managing a sexually transmitted infection. Sexual and Relationship Therapy, 20(1), 51-63.
RIVM (2017) Report on Sexually transmitted infections in the Netherlands in 2016. Retrieved from:
http://rivm.nl/Documenten_en_publicaties/Wetenschappelijk/Rapporten/2017/Juni/Se xually_transmitted_infections_including_HIV_in_the_Netherlands_in_2016
Rosenstock, I. M., Strecher, V. J., & Becker, M. H. (1988). Social learning theory and the health belief model. Health education quarterly, 15(2), 175-183.
Schmälzle, R., Renner, B., & Schupp, H. T. (2017). Health Risk Perception and Risk Communication. Policy Insights from the Behavioral and Brain Sciences, 4(2), 163-169.
Sensoa (2017). Symptomen van soa. Retrieved from:
https://www.seksualiteit.be/soas/symptomen-van-soa
Sheeran, P., Harris, P. R., & Epton, T. (2014). Does heightening risk appraisals change people’s intentions and behavior? A meta-analysis of experimental studies. Psychological bulletin, 140(2), 511.
Slovic, P., Monahan, J., & MacGregor, D. G. (2000). Violence Risk Assessment and Risk Communication: The Effects of Using Actual Cases, Providing Instruction, and Employing Probability Versus Frequency Formats. Law and Human Behavior, 24(3), 271-296.
Soa Aids Nederland (2017a). Soa’s. Retrieved from: https://www.soaaids.nl/nl/soas Soa Aids Nederland (2017b). Jongeren beginnen later aan seks. Retrieved from:
van Empelen, P., & Kok, G. (2006). Condom use in steady and casual sexual relationships: Planning, preparation and willingness to take risks among adolescents. Psychology & Health, 21(2), 165-181.
Veland, H., & Aven, T. (2013). Risk communication in the light of different risk perspectives. Reliability Engineering & System Safety, 110, 34-40.
Visschers, V. H. M., Meertens, R. M., Passchier, W. W. F., & De Vries, N. N. K. (2009). Probability Information in Risk Communication: A Review of the Research Literature. Risk Analysis, 29(2), 267-287.
Visschers, V. H. M., Wiedemann, P. M., Gutscher, H., Kurzenhäuser, S., Seidl, R., Jardine, C. G., & Timmermans, D. R. M. (2012). Affect-inducing risk communication: current knowledge and future directions.Journal of risk research,15(3), 257-271.
Weinstein, N. D. (1982). Unrealistic optimism about the susceptibility to health problems. Journal of Behavioral Medicine, 5, 441–460.
Weinstein, N. D. (2000). Perceived probability, perceived severity, and health-protective behavior. Health Psychology, 19(1), 65 - 74.
WHO (2017). Sexually transmitted diseases in 2017. Retrieved from: http://who.int/reproductivehealth/topics/rtis/en/
Wigfield, A., & Cambria, J. (2010). Students’ achievement values, goal orientations, and interest: Definitions, development, and relations to achievement outcomes. Developmental Review, 30(1), 1-35.
Wigfield, A., & Eccles, J. S. (2000). Expectancy–value theory of achievement motivation.Contemporary educational psychology, 25(1), 68-81.
Windschitl, P. D., & Wells, G. L. (1996). Measuring psychological uncertainty: Verbal versus numeric methods. Journal of Experimental Psychology: Applied, 2(4), 343.
Witte, K. (1992). Putting the fear back into fear appeals: The extended parallel process model.Communications Monographs,59(4), 329-349.
Witte, K., & Allen, M. (2000). A Meta-Analysis of Fear Appeals: Implications for Effective Public Health Campaigns. Health Education & Behavior, 27(5), 591-615
Witte, K., Berkowitz, J. M., Cameron, K. A., & McKeon, J. K. (1998). Preventing the Spread of Genital Warts: Using Fear Appeals to Promote Self-Protective Behaviors. Health Education & Behavior, 25(5), 571-585.
Wogalter, M. S., Young, S. L., Brelsford, J. W., & Barlow, T. (1999). The Relative Contributions of Injury Severity and Likelihood Information on Hazard-Risk Judgments and Warning Compliance. Journal of Safety Research, 30(3), 151-162. Wolfers, M., de Zwart, O., & Kok, G. (2011). Adolescents in The Netherlands Underestimate
Risk for Sexually Transmitted Infections and Deny the Need for Sexually Transmitted Infection Testing. AIDS Patient Care and STDs, 25(5), 311-319.
Appendix A
Study Design Table 1
Factorial design
Severe STI Consequences
STI Probability Low (X1) High (X2)
Low (Y1) X1Y1 X2Y1
