Educating marginalised adolescents about teenage pregnancy: How effective is a health-related photo story?

Photo stories – small booklets that tell a dramatic story through photos and captions – can enhance health education for disadvantaged groups. In an experiment involving 174 South African marginalised teenagers, a photo story about teenage pregnancy outperformed a traditional brochure and a control condition in terms of self-efficacy, attitudes and intentions to seek support. Participants also clearly preferred the photo story, especially the girls. The findings suggest that well-designed photo stories can be valuable tools for promoting positive attitudes and health behaviours.

Every year, approximately 21 million girls aged 15–19 years in developing regions become pregnant, of whom 12 million give birth. Most of these teenage pregnancies are unintended. Moreover, they are inclined to be higher among marginalised girls, particularly those with lower education or of low economic status. Teenage pregnancy, or adolescent pregnancy, presents dire health and socioeconomic consequences, not only for the women but also for the societies in which they are located. In lower middle-income countries, such as South Africa, sexual violence is one of the main causes of such pregnancies. Girls need to be made aware of this, and need to learn about ways to protect themselves and to access support (Kumar Jha et al., 2025; World Health Organization (WHO), 2024).

South Africa has one of the highest rates of teenage pregnancy in the world, with nearly one in four adolescents falling pregnant before the age of 20 (South African Government, 2024) and one in seven mothers in South Africa being an adolescent (Van Niekerk et al., 2024).

To address this problem, awareness campaigns about teenage pregnancy have been developed, with the government launching a pregnancy awareness week every year in February (South African Government, 2025). Despite these and other health education efforts to reduce teenage pregnancy rates, the problem appears to be growing. For example, the National Department of Social Development expressed concern that on New Year’s day 2026, the recorded number of 160 teenage births doubled the number recorded on New Year’s day 2025 (Magubane, 2026). Given the seriousness of the problem of teenage pregnancies, it is worth exploring alternative communication strategies to tackle the problem of teenage pregnancies in South Africa.

One way to improve health education in South Africa may be developing and distributing photo stories or fotonovelas: booklets in which photos and captions tell a dramatic story centred around a health theme. Photo stories have been deployed as health education tools for some 20 years now ^[1]. They have been used to educate and create awareness around an array of health issues, e.g. diabetes, hepatitis B, depression, dementia and human papillomavirus. Photo stories addressing health topics have primarily been designed for populations with limited literacy and health literacy skills (Cabassa et al., 2012; Unger et al., 2013). Koops van’t Jagt et al. (2018) suggest that these materials may also benefit individuals with higher levels of health literacy, making distribution easier. Their study on a Dutch diabetes photo story demonstrated that regardless of participants’ health literacy level, readers of the photo story scored higher on a knowledge assessment compared to readers of a conventional brochure and participants in a control condition.

According to international studies conducted on photo stories, a positive effect of photo stories may be an increase in knowledge about the health topic covered. In the United States, Unger et al. (2009) found that reading Sweet temptations/Tentaciones Dulces, a bilingual photo story about a family dealing with diabetes, led to a significant improvement in overall knowledge about diabetes. Valle et al. (2006) report a positive effect of a photo story about Alzheimer’s disease on older readers’ dementia knowledge. In the Netherlands, Koops van ’t Jagt et al. (2018) found, as mentioned above, that readers of a photo story on diabetes outperformed others on a knowledge test. In South Africa, James et al. (2005) developed Laduma (“It thunders” – a shout to celebrate a goal scored in soccer), a photo story that warns about sexually transmitted infections (STIs). Reading Laduma led to an increase in general knowledge about this topic, not only immediately but also six weeks later, and that these gains were greater than in a control group.

Also in South Africa, Davis and Jansen (2020) investigated the effects of Spyt kom te laat (Afrikaans for Regret comes too late) on the harmful effects of crystal meth use. Readers of the photo story performed better than participants in a control group in terms of their knowledge about this hard drug. A follow-up study found that older adults with low health literacy learned more from the crystal meth photo story than from the traditional brochure about the same topic, while older adults with high health literacy learned more from the brochure than from the photo story (Davis and Jansen, 2021). Davis and Jansen (2024) investigated the effects of two versions of Gertie se Groot Geheim (Afrikaans for Gertie’s Big Secret). Both versions of this photo story on diabetes resulted in more overall diabetes knowledge than a control condition. Here, younger adults (25–49) gained more knowledge when the main character was closer to their own age, though this effect was not seen in older readers.

As far as the effects on health-related behavioural attitudes or intentions are concerned, research into photo stories has not yet yielded a clear picture. For example, Koops van ’t Jagt et al. (2018) reported that their diabetes photo story did not outperform a brochure in strengthening intentions to engage in healthier behaviour. Unger et al. (2013) found that Secret Feelings, a photo story about depression, performed no better than a text pamphlet in reducing stigma or increasing the intention to seek help for depression. A study by Hernandez and Organista (2013) on the effects of the same photo story provides a more promising picture, showing reductions in stigma and marginal improvements in treatment-seeking intentions among Latino readers, though this comparison was made against a control group rather than a brochure.

South African studies paint a somewhat different picture. James et al. (2005) found that Laduma increased both positive attitudes towards condom use and future intentions to use condoms compared to a control group. Davis and Jansen (2020, 2021) found that their photo story intended for crystal meth prevention produced higher intentions to engage in interpersonal communication (talking to family or friends about meth use) compared to the brochure (although not the control group). In their 2021 study, the photo story produced higher attitudes to speak with family or friends about meth use when compared to the control condition (although not the brochure). In Davis and Jansen’s study (2024) on a diabetes photo story, limited but meaningful effects were also observed for attitudes and behavioural intentions. When compared to a control condition, positive effects for both attitudes and intentions were reported (e.g. willingness to speak to healthcare providers and intentions to exercise and eat more vegetables).

An important requisite for any communication message to have effect is the audience’s attention and continued engagement. If people do not want to read a health message, its content and quality become irrelevant (Davis and Jansen, 2021, drawing on McGuire’s input–output framework, 2013).To our knowledge, however, Davis and Jansen (2020, 2021) are the only researchers who have so far systematically examined readers’ preference for a photo story compared to a more health traditional brochure. In their 2020 study, they found that nearly twice as many participants preferred the photo story over a brochure, particularly younger and older age groups and those with lower educational attainment. By contrast, individuals with higher education levels showed no clear preference. In their follow-up study, Davis and Jansen (2021) again found a strong and statistically significant preference for the photo story across age groups and health literacy levels.

In photo stories, the characters talk directly to each other. The reader follows their conversations, where they share their opinions and together deliver the health message. In this sense, unlike other printed health materials, photo stories are like video health messages because they both show people talking and interacting. In a study on the effects of videos aimed at preventing teenage pregnancies, González-Cabrera and Igartua (2018) distinguish between “testimonial” stories and “dialogic” stories in health education videos. In the testimonial format, a narrator conveys the health message by addressing the audience with a story told in the first or third person. In the dialogic format, there are interactive characters who communicate directly with each other, and the audience witnesses how these characters express their opinions and jointly convey the health message. A successful African example of health-related stories in dialogic format is a music video distributed in Uganda, in which the health message was conveyed through a mix of images and songs. This video produced according to principles of Entertainment-Education (E–E; see Cabassa et al., 2012) demonstrated potential as an innovative medium for health promotion among African adolescents. Over 12 months, it was viewed 36,651 times on YouTube and generated more than 200,000 social media posts (Macnab and Mukisa, 2019; Macnab et al., 2024) ^[2].

To our knowledge, the distinction between a testimonial and a dialogic format has not yet been recognised in the research into printed health communication. In this field, there is a strong emphasis on the testimonial format, while printed health communication in dialogic form is still largely uncharted research territory. In a 2016 overview of characteristics of narrative interventions and effects in health-related persuasion research, De Graaf et al. assert that only a few narratives contain a health message in dialogues between characters, adding that in these studies, there was not “sufficient variation to draw conclusions about a pattern regarding their effectiveness” (2016, p. 95). As far as we know, the situation has not changed appreciably since the publication of this overview. Research on printed narratives in health education has largely concentrated on testimonial formats, and the effects of dialogic printed formats, such as photo stories, remain underexplored. The present study helps to address this gap by focusing on dialogic printed health education targeting marginalised young people in South Africa.

As shown above, photo stories can be part of the solution to various health education problems. The effects observed among young people in South Africa suggest that this medium may also be valuable in addressing teenage pregnancy. To our knowledge no previous studies, in South Africa or elsewhere have explored the potential of photo stories in combatting this specific health problem. In this study, we investigated this by testing three hypotheses, focusing on knowledge gain, behavioural attitudes and intentions, and readers’ preference.

First, as already discussed, studies from different countries including South Africa have found that readers of photo stories outperform readers of traditional brochures and participants in control groups in terms of health topic knowledge (Davis and Jansen, 2020, 2021, 2024; James et al., 2005; Koops van ’t Jagt et al., 2018; Unger et al., 2009; Valle et al., 2006). This leads the expectation that a new South African photo story about teenage pregnancy will positively affect the readers’ knowledge about this topic. See H1:

Second, previous international research, as discussed above, suggests that photo stories rarely exceed brochures in changing attitudes or intentions. In South Africa, however, photo stories have been found to be slightly more effective in this respect. This leads to the expectation that a new South African photo story could also have a positive effect on behavioural attitudes and intentions, as well as on underlying determinants such as perceived vulnerability and perceived self-efficacy, based on the EORM (Entertainment Overcoming Resistance Model, Moyer-Gusé, 2008). See H2:

Third, based on outcomes of earlier photo story studies in South Africa, we expect that a new South African photo story will be preferred over a traditional health brochure. See H3:

A randomised controlled trial (RCT) with three groups of girls and boys was conducted to assess the impact of different message formats in one of three conditions: reading a photo story, reading a traditional brochure or receiving no message (control group). The study measured differences in knowledge level, perceived vulnerability, perceived self-efficacy, behavioural attitudes and intentions. In addition, readers’ preferences for the photo story or the traditional brochure as a source of information about teenage pregnancy were examined.

A total of 180 participants were targeted for the three conditions: 60 per condition. Recruiters were instructed to ensure equal gender representation for each condition. Given that questionnaires were tailored for each gender (see Section 2.2.2), it effectively meant that the study had six different groups.

Prior to the main study, a pilot study (n = 20) was conducted, in which data from all six groups were collected simultaneously and at the same location. Feedback from this pilot study showed that it was logistically difficult to supervise the six groups simultaneously, especially since adolescents were involved. Because we did not want to take the risk of not being able to meet the experimental requirements, we decided to hold separate sessions for each of the three conditions, starting at three different times on the same day.

To prevent possible systematic differences between the various conditions, field workers were instructed as part of the recruitment process to ensure that an equal mix of participants were included in each of the three sessions in terms of age (12–17 years) and school representation. Totals were as follows: all conditions: n = 174 (96 girls and 78 boys); photo story group: n = 60 (26 boys and 34 girls); traditional brochure group: n = 56 (31 boys and 25 girls); control group: n = 58 (21 boys and 37 girls); also see Sections 2.3 and 3.1.

A photo story about teenage pregnancy, entitled “It happened to me”, was developed for this study, using an E–E approach involving a six-step production process (Cabassa et al., 2012). This document was developed and workshopped in collaboration with the target audience and other community members and representatives from NGOs. The cultural-centric narrative of the photo story was primarily shaped by constructs from Larkey and Hecht’s model of culture-centric narratives. According to their framework, elements such as culturally authentic characters and plotlines enhance audience engagement and influence the effectiveness of health messages (Larkey and Hecht, 2010). Earlier research by Garro (1994) supports this idea, demonstrating that shared cultural models among participants were crucial for interpreting their narratives about the illness.

The model of culture-centric narratives was operationalised as follows. All the scenes in the photo story were shot on location in a typical informal settlement where teenage pregnancy rates are known to be high. Actors from the area were recruited to appear in the photo story. Locations included a local school, a community member’s actual house and a real tavern in the area. Following the model of culture-centric narratives resulted in the development of a “peer to peer” narrative. The WHO Health Promoting School model (HPS) emphasises the likelihood of peer-driven materials being more effective and more likely to have a lasting impact (Macnab, 2013; WHO, 2020). The photo story was developed in English, so that all intended readers could understand it (South Africa has 11 official spoken languages, of which English is understood by most adolescents). See Figure 1. For the complete photo story, see “Supplementary_material_appendix_1_photo story”.

In order to compare the photo story with an existing teenage pregnancy document for South African youth, we searched the website of the South African Department of Health (www.health.gov.za) and related sites, but without success. We also inquired at local clinics and health institutions in Gauteng province about the availability of documents on teenage pregnancy such as leaflets, flyers or brochures. The responses from health workers indicated that no printed copies of such documents were readily available in public hospitals. For this reason, we decided to develop a traditional type of brochure ourselves, entitled “Teenage Pregnancies.” The information in this four-panel fold-out brochure closely matched that of the photo story, and great care was taken to ensure a professional look and layout. Careful attention was paid to ensuring that the information would be easily accessible. Principles of plain language were applied. Short paragraphs, short sentences and words of one or two syllables were used, and potentially difficult terms were avoided as much as possible. According to an online calculation, this resulted in a Flesch Reading Ease score of 64.9; Reading Level 8 and 9 (Plain English) ^[3].

To ensure that the brochure would match the format of traditional brochures often used by national and provincial health departments in South Africa, 6 health professionals from 3 hospitals in Johannesburg were approached: 4 males and 2 females; 3 medical doctors and 3 medical students. All health professionals were asked to assess whether the brochure was suitable for its intended purpose. They all considered the brochure to be typical of traditional brochures used for health education in South Africa. According to their further responses, the brochure would be effective for disseminating information about teenage pregnancy, it contained relevant content, had a clear message, had a clear layout and was suitable for use by the health service. See Figure 2. For the complete brochure see “Supplementary_material_appendix_2_brochure”.

In the pilot study referred to above, self-administered questionnaires for all conditions were tested with 20 participants from the target group (also see Section 2.1). Any ambiguities that emerged were corrected. The questionnaires included items about demographic characteristics, knowledge, perceived vulnerability, perceived self-efficacy, behavioural attitudes and behavioural intentions. Parts of the questionnaires had to be tailored for each gender. To measure perceived vulnerability, one of the questions for girls was, for instance: “If you have unprotected sex once in the near future, how likely is it that you will get pregnant? For boys this question was”: “If you have unprotected sex with a girl once in the near future, how likely is it that she will get pregnant?”

The study also investigated whether readers preferred the photo story or the traditional brochure as a source of information about teenage pregnancy (see Section 2.5.7).

The study was conducted in the previously disadvantaged informal settlements of Gugulethu and Nyanga, in the vicinity of Cape Town in South Africa’s Western Cape province. The participants (n = 174) were predominantly IsiXhosa-speaking boys and girls of high school age (typically aged 12–17). They were recruited for the study with the help of a local NGO working with youths in the area.

We decided to include not only girls but also boys in the study, in view of various studies that show the positive outcomes of male partner involvement. It improves pregnancy partner healthcare utilisation (Alemi et al., 2021). Furthermore, involvement in pregnant partner healthcare increases if boys have been exposed to information on male involvement in antenatal health (Degefa et al., 2024). Additionally, in most cultures, also in South Africa, women are neither the primary nor the only decision-makers about their health (Yargawa and Leonardi-Bee, 2015).

Given that the participants were aged between 12 and 17 years, written consent was requested and obtained from their parents and guardians; verbal consent was obtained from the participants themselves.

This study was approved by the University of Johannesburg, Faculty of Humanities Research Ethics Committee. All ethical guidelines were followed.

Data were collected on a Saturday at a local school by the researchers with the help of field workers who were briefed about the format of the experiment before it was conducted. During all sessions, a fieldworker explained what the research was about.

Depending on the condition, each participant was given either a photo story and accompanying questionnaire, or a traditional brochure and accompanying questionnaire, or only a questionnaire (control group). All three groups first answered demographic questions (part 1 of the questionnaire), followed by knowledge questions related to teenage pregnancy, behavioural attitudes, behavioural intentions, perceived vulnerability, perceived self-efficacy (part 2 of the questionnaire) and health message preference (part 3 of the questionnaire). After having completed part 1, participants in the experimental conditions (photo story or traditional brochure) were asked to read the health document in the format that corresponded to their condition. They were then instructed to return the health document to the researchers or field workers. After this, they completed and returned part 2 of the questionnaire. Next, they were asked to read the health message in the other format, and to answer questions indicating their evaluation of the two formats (part 3). Participants in the control group were asked to first complete part 1 and part 2 of the questionnaire, and then to read the photo story and the traditional brochure in the order they preferred. After this, they completed the evaluation questions about the two message formats (part 3). All participants were offered refreshments afterwards. No financial incentives were given.

Participants were asked to indicate their gender, home language, school, grade and actual age.

Knowledge level related to teenage pregnancy was measured with eight statements that could be either true or false. Factual information about the psychosocial challenges of teenage pregnancy in South Africa was sourced from Ntshayintshayi et al. (2022), and information about the signs and symptoms of pregnancy from Better Health Channel (2024). Each answer was scored as 1 for “correct”, 2 for “incorrect” (no answer was also regarded as “incorrect”) and 3 for “don’t know”.

True statements were the following: “1. Pregnant women may experience weight gain”, “2. Pregnancy can cause you to feel tired often”, “3. Symptoms of being pregnant include a missed menstrual period”, “6. Teenage mothers may face stigma about being pregnant so young”, and “7. Teenage mothers may run into financial problems”. False statements were “4. Hearing problems is one of the symptoms of pregnancy”, “5. Having sex once cannot cause pregnancy”, and “8. Teenage mothers are on their own without support”. All answers could be found in both the photo story and the traditional brochure.

To measure “perceived vulnerability”, 3 items were used, based on Koops van ’t Jagt et al. (2018), referring to Moyer-Gusé and Nabi (2010) and Gibbons et al. (2003). Answers could be given on 5-point scales (1 = very small and 5 = very big).

The items were Item 1, female version: “In general, how likely is it that you will get pregnant in the near future?”; male version: “In general, how likely is it that a girl you have sex with will get pregnant in the near future?”; Item 2, female version: “If you have unprotected sex once in the near future, how likely is it that you will get pregnant?”; male version: “If you have unprotected sex with a girl once in the near future, how likely is it that she will get pregnant?”; and Item 3, female version: “If you have unprotected sex more than once in the near future, how likely is it that you will get pregnant?”; male version: “If you have unprotected sex with a girl more than once in the near future, how likely is it that she will get pregnant?”

Reliability analysis showed that it was not possible to combine the reactions to these items into one reliable scale: Cronbach’s alpha = 0.34. Leaving out one of the items did not lead to an acceptable Cronbach’s alpha either. Based on these outcomes, it was decided not to use the variable “perceived vulnerability” for further statistical analyses.

To measure aspects of “perceived self-efficacy”, three items were used, following Frei et al. (2009). Again, 5-point scales were used (1 = strongly disagree and 5 = strongly agree).

Item 1 was about waiting to have sex: female version: “If a guy I really like wants to have sex with me, I think I can say no and wait until I am older”; male version: “If I really like a young girl and want to have sex with her, I think I can wait until she is older.” Item 2 was about birth control: female version: “If a guy I really like wants to have sex with me, I think I can insist on using a condom or other birth control”; male version: “If I really like a young girl and want to have sex with her, I think I can insist on using a condom or other birth control.” Item 3 was about looking for support: female version: “If I have sex with a guy and I fall pregnant, I think I can look for support for adolescent mothers”; male version: “If I do have sex with a young girl and she falls pregnant, I think I can encourage her to look for support for adolescent mothers.”

The items used to measure aspects of “perceived self-efficacy” differed too much in content to combine them meaningfully. Cronbach’s alphas were therefore not calculated.

To measure “behavioural attitudes”, three items were developed, based on Frei et al. (2009). Again, 5-point scales were used (1 = strongly disagree and 5 = strongly agree).

Item 1 was about waiting to have sex: female version: “If a young girl really likes a guy and he wants to have sex with her, it is a good thing if they wait until she is older”; male version: “If a guy really likes a young girl and wants to have sex with her, it is a good thing if they wait until she is older.” Item 2 was about birth control: female version: “If a young girl really likes a guy and he wants to have sex with her, it is a good thing if she insists on using a condom or other birth control”; male version: “If a guy really likes a young girl and wants to have sex with her, it is a good thing if he insists on using a condom or other birth control.” Item 3 was about looking for support. Female version: “If a guy and a young girl have sex and she falls pregnant, it is a good thing if she will look for support for adolescent mothers”; male version: “If a guy and a young girl have sex and she falls pregnant, it is a good thing if the guy encourages her to look for support for adolescent mothers.”

Because of content differences between the three attitudes measured, all behavioural attitude items were analysed separately, and Cronbach’s alphas were not calculated.

To measure “behavioural intentions”, also three items were developed following Frei et al. (2009), and 5-point scales were used (1 = strongly disagree and 5 = strongly agree).

Item 1 was about waiting to have sex: female version: “If a guy I really like wants to have sex with me, I will definitely wait until I am older”; male version: “If I really like a young girl and want to have sex with her, I will definitely wait until we are older.” Item 2 was about birth control: female version: “If a guy I really like wants to have sex with me, I will definitely insist on using a condom or other birth control”; male version: “If I really like a young girl and want to have sex with her, I will definitely insist on using a condom or other birth control.” Item 3 was about looking for support: female version: “If I have sex with a guy and I fall pregnant, I will definitely look for support for adolescent mothers”; male version: “If I do have sex with a young girl and she falls pregnant, I will definitely encourage her to look for support for adolescent mothers.”

Because of content differences between the three intentions measured, all behavioural intention items were analysed separately, and Cronbach’s alphas were not calculated.

For measuring message preference, seven evaluation questions were asked, using a 5-point scale (1 = strongly disagree and 5 = strongly agree). The questions, developed by the researchers, were: “1. Reading the brochure takes more time than reading the photo story”; “2. Reading the brochure takes more effort than reading the photo story”; “3. The photo story provides more useful information about teenage pregnancy than the brochure”; “4. The brochure is more fun than the photo story”, “5. I would rather read the photo story than the brochure”; “6. It is a good idea to use the photo story to educate young people about teenage pregnancy”; and “7. It is a good idea to use the brochure to educate young people about teenage pregnancy.”

To answer the research questions, descriptive statistics, t-tests and univariate and multivariate analyses of variance (alpha always set at 0.05) were performed, using IBM SPSS version 30.0.

In total 180 participants completed a questionnaire. There were six participants who did not reveal their gender, or who provided information about their gender that did not match the version of the questionnaire that was given to them. The data from these six participants were disregarded in further statistical analyses. For the distribution of the 174 other participants (96 boys and 78 girls) over the two experimental conditions and the control condition, see Table 1, which also provides information about the participants’ home language, school grade and age.

The first hypothesis proposed that, compared to both readers of a traditional brochure and a control group, readers of a photo story about teenage pregnancy would show more topic-related knowledge. Table 2 shows the proportions of correct, incorrect and “don’t know” answers for the different conditions and genders (means and standard deviations).

Outcomes of univariate analyses of variance showed the following:

1. For correct answers, there was no statistically significant effect of condition, neither for the complete group (p = 0.40) nor for the boys (p = 0.50) nor for the girls (p = 0.14). For the girls, however, post hoc tests revealed one difference bordering significance (p = 0.06), the difference between the photo story condition (m = 0.66; sd = 0.22) and the brochure condition (m = 0.56; sd = 0.20).

2. For incorrect answers, there was no statistically significant effect of condition for the complete group (p = 0.08), nor for the boys (p = 0.96). For the girls, the effect of condition was significant F(2,93) = 3.86 (p < 0.05). Post hoc tests showed a significant difference (p < 0.01) between the photo story condition (m = 0.16; sd = 0.16) and the brochure condition (m = 0.28; sd = 0.19). Other differences between conditions were not significant (p > 0.10).

3. For “don’t know” answers, there was a statistically significant effect of condition for the complete group of participants: F(2,171) = 6.01 (p < 0.01). Post hoc tests showed a significant difference (p < 0.01) between the photo story condition (m = 0.18; sd = 0.19) and the brochure condition (m = 0.08; sd = 0.12). The difference between the control condition (m = 0.18; sd = 0.20) and the brochure condition was also significant (p < 0.01). There was no significant difference between the photo story condition and the control condition (p = 0.86). For the boys, there was no statistically significant effect of condition (p = 0.11). For the girls, the effect of condition was significant F(2,93) = 3.76 (p < 0.05). Post hoc tests showed a significant difference (p < 0.05) between the photo story condition (m = 0.18; sd = 0.20) and the brochure condition (m = 0.09; sd = 0.10). The difference between the control condition (m = 0.22; sd = 0.22) and the brochure condition was also significant (p < 0.01). There was no significant difference between the photo story condition and the control condition (p = 0.48).

4. Analyses of variance were also performed to test the effect of condition on the difference between the mean scores for “incorrect” and “don’t know” answers. For the complete group of participants, a significant effect was found: F(2,171) = 6.30 (p < 0.01). Post hoc tests showed a significant difference (p < 0.001) between the photo story condition (m = 0.03; sd = 0.28) and the brochure condition (m = 0.20; sd = 0.19). The difference between the control condition (m = 0.06; sd = 0.32) and the brochure condition was also significant (p < 0.01). There was no significant difference between the photo story condition and the control condition (p = 0.50). For the boys, there was no statistically significant effect of condition (p = 0.31). For the girls, the effect of condition was significant F(2,93) = 3.89 (p < 0.05). Post hoc tests showed a significant difference (p < 0.01) between the photo story condition (m = −0.02; sd = 0.29) and the brochure condition (m = 0.20; sd = 0.19). The difference between the control condition (m = 0.01; sd = 0.32) and the brochure condition was also significant (p < 0.05). There was no significant difference between the photo story condition and the control condition (p = 0.60).

In summary, there was only limited support for H1. Girls in particular benefitted from reading the photo story in one respect: they agreed significantly less often with factually incorrect statements than girls who had read the brochure. Condition had no significant effect on the percentage of correct answers. This applied to both the complete group of participants and the subgroups of boys and girls. For both the complete group and the subgroup of girls, a significant effect was found on the difference between the percentage of incorrect answers and the percentage of “don’t know” answers. Compared to reading the brochure, reading the photo story led to a greater degree of uncertainty about statements that were in fact incorrect.

According to the second hypothesis, readers of a photo story about teenage pregnancy would report higher levels of perceived vulnerability, perceived self-efficacy, behavioural attitudes and behavioural intention, compared to both readers of a traditional brochure and a control group.

Multivariate analyses of variance for 9 items, measuring aspects of perceived self-efficacy, behavioural attitudes and behavioural intention revealed the outcomes shown in Table 3. The effects on perceived vulnerability could not be tested, due to insufficient measurement reliability (see Section 2.5.3).

In view of the multivariate effect for girls bordering on significance (p = 0.06), follow-up univariate analyses were performed for this subgroup, revealing statistically significant effects for 4 of the 9 items. See Table 4 for mean scores and standard deviations on 5-point scales (1 = strongly disagree and 5 = strongly agree).

1. For Self-efficacy (item 2) the effect of condition was significant: F(2,92) = 3.12 (p < 0.05). Post hoc tests showed a significant difference (p < 0.05) between the photo story condition (m = 3.56; sd = 1.16) and the control condition (m = 2.97; sd = 1.30). The difference between the brochure condition (m = 3.67; sd = 1.13) and the control condition was also significant (p < 0.05). There was no significant difference between the photo story condition and the brochure condition (p = 0.74).

2. For Self-efficacy (item 3) there also was a significant effect of condition: F(2,93) = 3.31(p < 0.05). Post hoc tests showed a significant difference (p < 0.05) between the photo story condition (m = 3.74; sd = 0.99) and the control condition (m = 3.03; sd = 1.28). Other differences between conditions were not significant (p > 0.10).

3. For Behavioural attitude (item 3) the effect of condition also was significant: F(2,92) = 4.64 (p < 0.05). Post hoc tests showed a significant difference (p < 0.01) between the photo story condition (m = 3.62; sd = 0.99) and the control condition (m = 2.76; sd = 1.42). The difference between the photo story condition and the brochure condition (m = 2.83; sd = 1.40) was also significant (p < 0.05). There was no significant difference between the brochure condition and the control condition (p = 0.82).

4. For Behavioural intention (item 3) there also was a significant effect of condition: F(2,92) = 3.60 (p < 0.05). Post hoc tests showed a significant difference (p < 0.05) between the photo story condition (m = 3.62; sd = 1.28) and the control condition (m = 2.89; sd = 1.35). The difference between the brochure condition (m = 3.62; sd = 1.21) and the control condition was also significant (p < 0.05). There was no significant difference between the photo story condition and the brochure condition (p = 0.98).

In brief, H2 was partially supported, but only for girls. In this subgroup, both readers of the photo story and the brochure reported a significantly higher level of self-efficacy regarding birth control than participants in the control condition. For all three items with regard to “looking for support” (perceived self-efficacy, behavioural attitude and behavioural intention), the scores in the photo story condition were significantly higher than in the control condition. On the behavioural attitude item about “looking for support”, readers of the photo story also scored significantly higher than readers of the traditional brochure. Readers of the photo story also scored significantly higher on the behavioural attitude item regarding “looking for support” than readers of the traditional brochure.

The third hypothesis proposed that readers would prefer a photo story about teenage pregnancy to a traditional brochure about the same topic.

To test this hypothesis, two-sided t-tests were performed, determining whether the mean scores for the evaluation items differed significantly from the midpoint (3) of the 5-point scales used (1 = strongly disagree and 5 = strongly agree). For the outcomes, see Table 5.

Table 5 shows that readers preferred the photo story to the brochure (item 5). Compared to the brochure, the photo story took the same amount of time but required less effort (items 1 and 2), indicating that the photo story was found easier to understand. In the opinion of the participants, the photo story was neither funnier nor less funny than the brochure (item 4), but it offered more useful information (item 3). Both the photo story and the brochure were considered useful tools for educating young people about teenage pregnancy (items 6 and 7). A two-sided t-test showed that the difference between the scores for items 6 and 7 was not significant: t(170) = 1.47 (p = 0.14).

To identify possible main or interaction effects of condition and gender on the evaluation items, univariate analyses of variance were performed. No significant main effect of condition was found. There were two significant main effects of gender. On evaluation item 6, girls (m = 4.18; sd = 1.02) scored higher than boys (m = 3.63; sd = 1.25): F(1,169) = 9.91 (p < 0.001). Furthermore, there was a significant effect of gender on the difference between the scores for items 6 and 7: F(1,169 = 7.53 (p < 0.01). For girls the difference between the scores for items 6 and 7 was positive (0.43); for boys it was negative (−0.17). Further analyses showed that for girls this difference significantly deviated from 0; t(94) = 2.85 (p < 0.01); for boys it did not: t(75) = −1.10 (p = 0.27).

In summary, the significant effects found showed that participants overall preferred the photo story to the brochure. They felt that reading the photo story took less effort than reading the brochure, while the photo story provided more useful information. Both the photo story and the brochure were considered suitable for educating young people about teenage pregnancy. However, girls were significantly more positive about the use of the photo stories than about the brochure as a tool for educating young people about teenage pregnancy.

In this study, a photo story on teenage pregnancy worked better than a traditional brochure including the same information. This applies to a limited extent to increasing relevant knowledge, but the effects on broader determinants of behaviour were more clearly visible. The most striking outcome was the target group’s clear preference for the photo story. It was found to be easier to understand and more useful. Girls in particular were more positive about the use of photo stories than about the brochure to educate young people about teenage pregnancy.

Compared to previous photo story studies, both in South Africa and elsewhere (Davis and Jansen, 2024; Koops van ’t Jagt et al., 2018; Unger et al., 2009), the knowledge effects found in this study were modest. The participants proved to understand most of the content of both the photo story and the brochure. Whereas condition had no significant effect on the percentage of correct answers, a favourable effect was observed in terms of the percentage of incorrect answers among girls in the photo story condition compared to the brochure condition.

While the knowledge effects were modest, the results regarding behavioural attitudes and intentions point to the potential of narrative formats to influence broader determinants of behaviour. As in previous studies, readers of the photo story, in this case girls in particular, reported positive attitudes and intentions towards the targeted health behaviour. The scores for self-efficacy with regard to birth control and “seeking support” indicate that both the brochure and the photo story caused girls to view teenage pregnancy differently. The scores for self-efficacy and behavioural attitudes with regard to “seeking support” show that in this respect the changes were greatest among the girls who had read the photo story.

These findings support the assumption that stories presented in a dialogic, realistic form can promote behavioural change, particularly in communities with limited resources.

The preferences of the readers also offer an important perspective. Participants often indicated that they found the photo story more engaging and easier to understand than the brochure. Girls in particular thought it was a good idea to use the photo story to educate young people about teenage pregnancy. This preference for narrative, visual storytelling is consistent with previous research emphasising the motivational appeal of entertainment-educational (E–E) materials (Davis and Jansen, 2024; Moyer-Gusé and Dale, 2017; Unger et al., 2009). Such preferences are relevant, as interest is an important prerequisite for any subsequent change in knowledge or attitude.

The observed gender differences – with stronger effects among girls than boys – warrant further investigation. One possibility relates to the subject matter of the health education materials compared. Information about teenage pregnancy may be more appealing to young girls than to young boys. In addition, the gender of the main character in the photo story may have promoted identification among girls in particular. Previous studies have shown that actual and perceived similarity between readers and main characters, and identification with a character, can influence the story’s effects on the reader (see, for example, Davis and Jansen, 2024; Koops van’t Jagt et al., 2018; Moyer-Gusé, 2008). Future research could further investigate how such mechanisms may influence the impact of narrative health education materials.

A key strength of the current study is that it focuses on a highly relevant social issue – teenage pregnancy – and on a marginalised target group in South Africa, that is underserved by conventional print education approaches. Furthermore, this study included both girls and boys, and examined not only the impact on knowledge but also on attitudes, intentions and preferences. It also was the first study in which both a photo story and a traditional brochure on teenage pregnancy were developed and empirically tested in an underprivileged community.

The involvement of young women and men in developing the photo story is an added strength of the study, given that youth engagement is often neglected in health promotion initiatives for that age group. Additionally, it may have contributed to the resonance of the photos with the participants, who were of a similar age group. These factors make this study an important step in understanding how printed dialogic storytelling material can be used to support health education and social change in such environments.

However, some limitations should be acknowledged. First, the timing of data collection differed slightly between the different conditions, which may have led to uncontrolled variability. Second, only one photo story and one traditional brochure were used, which means that the findings cannot be readily generalised to other topics, designs or populations. Third, the traditional brochure was developed in-house and was not existing education material, which may have limited the ecological validity. Fourth, the scores on items assessing perceived vulnerability could not be used in the statistical analyses. It proved impossible to combine the responses to these items into a single reliable scale. Fifth, behavioural attitudes and behavioural intentions were measured immediately after exposure to the education material. Although such immediate effects are informative, they do not necessarily lead to long-term behavioural change. A follow-up measurement at a later time, ideally with indicators of actual behaviour, would have provided a more comprehensive insight into the impact of the material. Future studies may include longitudinal designs to capture the sustainability of the effects.

The current findings emphasise the need for further research into explanations for differences in responses to narrative health messages on topics such as teenage pregnancy. The stronger effects among girls may reflect identification with the female protagonist, but other factors, such as the topic of the education tool, may also play a role. Future research should systematically investigate how narrative mechanisms, such as perceived similarity and identification, combined with topic relevance, influence the impact of health education tools. New studies could also explore variations on the photo story format and compare them with other types of narrative and non-narrative health education tools in different cultural and literacy contexts.

In addition to printed photo stories, online versions also deserve serious consideration. Even among marginalised young people in Southern Africa, smartphone ownership is now widespread, and digital, preferably interactive photo stories can be distributed on a large scale via this channel. The ability to comment on and ask questions about photo stories distributed online is expected to enhance the target group’s engagement. The resulting dialogue with young people can increase the reach of the photo story and reinforce its positive impact; compare the above example of the Ugandan music video (Macnab and Mukisa, 2019; Macnab et al., 2024).

It should be noted, however, that a large-scale study from 2023 analysing data from more than 500,000 women in 34 low-income and middle-income countries, mostly in sub-Saharan Africa, found that although mobile phone ownership is increasing rapidly, its distribution will remain uneven for some time. Inequalities in use of digital technology persist: mobile phone use is related to gender, socioeconomic status, technological proficiency and literacy (Sánchez-Páez et al. (2023, p. 1, p. 2). As Greenleaf et al. (2025) conclude from their study into mobile phone ownership among young adults in seven Southern African countries: “Considering the consistent association between key sociodemographic characteristics and phone ownership across countries, those employing mobile phones for youth health programmes may not reach the general youth population.” It therefore seems advisable to also continue to pay attention to printed photo stories, not least because understanding the characteristics that explain how they work can contribute to the development of successful online versions.

From a practical point of view, the results suggest that both traditional brochures and narrative formats can be valuable health education tools in an underprivileged community, provided they are well designed and tailored to the target audience. The findings also indicate that photo stories, by combining visual storytelling with recognisable characters and dialogues, are preferred by readers over health brochures and can be effective in promoting positive attitudes and intentions. Health educators may therefore consider incorporating photo stories or similar narrative formats into their communication toolbox.

This study showed that, particularly in an underprivileged community, a well-written brochure on a health topic such as teenage pregnancy can be effective, but that a well-designed photo story likely works even better. While this applies to a limited extent to increasing relevant knowledge, the effects on broader determinants of behaviour are more clearly visible. However, since interest in a message is a prerequisite for any subsequent change in knowledge or behaviour, perhaps the most important outcome is the clear preference of the target group for narrative, visual storytelling.

The supplementary material for this article can be found online at: For the complete photo story, see www.shorturl.at/EVv9J; for the complete brochure, see www.shorturl.at/AQU8A