Introduction

Globally, young people (i.e., those aged 10–24 years) are now reaching puberty earlier than in previous decades, and rates of early sexual debut are also rising or remain unchanged [1]. Ordinarily, simply engaging in premarital sexual intercourse is not a major health issue; however, it can be a precursor for risky sexual behavior later in life [2]. If no protection is used, premarital sex debut can result in serious adverse sexual health outcomes, such as an unwanted pregnancy, an abortion, regret, guilt, loss of self-respect, depression, loss of family support, substance abuse, and suicide [3]. Risk factors for premarital sex include being older; living alone or with friends; being in primary/elementary school or college; getting pocket money; having divorced or widowed parents [4, 5]; peer pressure; having strict parents; religiosity [6, 7]; watching pornographic material; substance abuse; being male [5, 8]; living in rural areas; and being in a romantic relationship [9]. Parent-child relationships; family environment; societal environment; cultural and traditional rules and values; economic conditions; and school environment are also some predictors of premarital sex debut [3].

In sub-Saharan Africa, efforts by national governments to control the spread of human immunodeficiency virus (HIV) among youth (i.e., those aged 15–24 years) have not yielded the desired results, as this group still accounts for two-thirds of all people living with HIV in this region [10]. Eswatini (formerly Swaziland), a small landlocked country in southern Africa with 1.1 million inhabitants, has not been spared this pandemic, as 20.9% of females and 4.2% of males aged 20–24 years are living with HIV [11], 94% of which was acquired through heterosexual intercourse [12]. Ironically, this occurs in a country where more than 90% of the population is Christian [13], a religion that strongly condemns premarital sex and regards it as immoral and sinful (fornication). In addition, Eswatini is one of the few countries in the world with a strong cultural heritage and highly conservative cultural norms and values, and traditions are highly safeguarded and widely practiced by a majority of the populace.

In Eswatini culture, premarital sex is strongly condemned, while chastity is encouraged and promoted through various traditional practices such as Umhlanga (Reed Dance), an annual national ceremony where unmarried girls who have no children assemble at the King’s palace to celebrate their chastity. Umcwasho, a tassel worn on the head by unmarried girls who have no children during a prescribed period (usually lasting 1–5 years) when a girl is not supposed to be ‘touched’ by a boy, is an example of a similar cultural practice. Males also have their own similar annual cultural practice called Lusekwane (a national sacred shrub), whereby all unmarried boys who have no children are expected to assemble at the King’s palace to receive the royal command to go and cut the sacred shrub during the annual Incwala ceremony [14]. These traditions are aimed at promoting chastity and therefore discouraging sex before marriage among the youth. However, despite the above-described religious rules and cultural customs, 54.5% of young women and 53.6% of young men aged 15–24 years reported having had sex [15], 3.5% of whom had sex before the age of 15 years [11]. Being cognizant of this problem, in 2014, the Eswatini government introduced a sex education curriculum in high schools across the country; however, children do not receive sexual education at home, as talking about sexuality issues is taboo in many communities and households [16]. This is because parents fear that it may promote promiscuity and may serve as a license for adolescents to practice sex [17].

In her seminal work in the 1960s, Baumrind [18–20] conceptualized three parenting style typologies: authoritative (high demandingness and high responsiveness), authoritarian (high demandingness and low responsiveness), indulgent/permissive (low demandingness and high responsiveness). She established that authoritative parenting produced assertive, self-reliant children, whereas authoritarian parenting produced discounted, withdrawn children. On the other hand, permissive parenting produced children with low self-control and low self-reliance. Later, Maccoby and Martin [21] conceptualized parenting styles along two dimensions: responsiveness (the extent to which parents are warm, supportive, sensitive, interested, non-coercive, and accustomed to their children’s needs) and demandingness (the degree to which parents demonstrate firm control, set maturity demands, provide supervision, enforce discipline, engage in direct confrontation, and establish performance or behavioural expectations for their children) [18, 19, 22]. Maccoby and Martin added a fourth parenting style: neglectful/uninvolved (low demandingness and low responsiveness) which is said to be associated with the worst child outcomes [21, 23].

In the early 90s, Darling and Steinberg proposed the contextual model of parenting style aimed at explaining the mechanism through which parenting practices influence child outcomes [24]. They defined parenting styles as constellations of attitudes that parents display towards their children that create emotional environments in which parents’ behaviors are expressed [24]. The model postulated that parenting styles are different from parenting practices in that they consist of parent-child interactions across a wide range of situations, yet parenting practices are situation-specific and goal-directed behaviors that parents use to socialize their children in a particular setting [24]. The model envisions parenting style as a contextual variable that moderates the relationship between specific parenting practices and specific developmental outcomes. It hypothesizes that parenting styles alter the efficacy of parents’ socialization efforts by buffering the effectiveness of particular practices, and by changing the child’s openness to socialization. Therefore, the model does not take parenting style as a developmental process, but rather a characteristic of the parent and a feature of the child’s social environment [24].

Research has shown that parenting styles have direct and indirect effects on children’s outcomes [25], whereby adolescents of authoritative parents who have positive parental relationships, healthy open communication, and perceived parental support, are said to be less likely to engage in sexual risk behaviors [26]. Low levels of perceived parental warmth and parental knowledge have also been found to predict sexual onset [27]. To the contrary, using single item measures for each parenting style, Huebner and Howell [28], did not find direct effects of parenting style on sexual risk-taking.

Of all the parenting styles, authoritative parenting has been found to be the most effective style to child-rearing, as it leads to children who are socially responsible, achievement-oriented, and competent, with good decision-making abilities [23, 29]. However, there is paucity of studies testing the contextual model of parenting style. The few studies that have examined the relationship between parental practices and child outcomes, but not specifically testing the contextual model, have found that this relationship is strongest for children of authoritative parents [29–31]. Unfortunately, these studies have not adequately addressed the mechanisms through which parenting styles influence youth sexual behaviors.

Additionally, most research on parenting styles and parent-child processes has been conducted in developed countries, including in many ethnic societies (e.g., in African-American communities) [32], but little has been done in the African region [33]. As a result, the mechanism through which parent practices and behaviors influence youth sexual behaviors is not well understood in this region, let alone in Eswatini. To try to understand these mechanisms, we adapted Darling and Steinberg’s [24] contextual model of parenting style and examined whether parenting styles moderated the relationship between parent-youth sexual risk communication (PYSRC) and premarital sexual debut among in-school youth in Eswatini. The contextual model was used as a foundation for formulating the research questions, as well as for the hypothesized directions in our conceptual framework (see S1 Fig). We hypothesized that parenting style would moderate the relationship between parent-youth sexual risk communication (parental practice) and engagement of the youth in premarital sex (child outcome).

Methods

Results

Background characteristics

As shown in Table 1, a majority (54.3%, n = 251) of participants were female, enrolled in urban schools (53.5%, n = 247), were unsure as to the number of friends who had already engaged in premarital sex (46.3%, n = 214), and 53% (n = 245) reported to have regular interactions with both of their parents. Nearly 36% (n = 166) of the participants had had premarital sex, a majority of whom did so at 17 years or older (54.8%, n = 91). In the bivariate analysis, participants who had engaged in premarital sexual significantly differed from those who had not by gender (p < 0.001), school location (p = 0.008), grade (p = 0.005), peer sexual activity (p < 0.001), most influential parent’s education level (p = 0.04), and age (p < 0.001). However, there were no statistically significant differences in the mean scores of PYSRC (p = 0.49 and of authoritativeness (p = 0.15) between the two groups (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of participants by premarital sexual debut.

https://doi.org/10.1371/journal.pone.0245590.t001

Multiple logistic regression results

Interpretation and reporting of the main effects in this study followed recommendations by Frazier and colleagues [53]. After adjusting for other covariates in step 4 (model 4), neither PYSRC (AOR = 1.02, 95% CI: 0.994, 1.039) nor authoritativeness (AOR = 0.82, 95% CI: 0.643, 1.050) significantly predicted youth premarital sexual debut, when authoritativeness score is conditioned at 0 in each case. Similarly, the interaction term (PYSRC x Authoritativeness) showed no statistically significant moderating effect on the association between PYSRC and premarital sexual debut (AOR = 1.01, 95% CI: 0.995, 1.022), as shown in Table 2. However, it is worth mentioning that in model 3 (i.e., after adding authoritativeness), PYSRC became statistically significant, (AOR = 1.02, 95% CI: 1.001, 1.042, p = 0.044)], but the borderline significance did not hold in model 4 when the interaction term was added (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Hierarchical logistic regression model testing moderation effects (N = 462).

https://doi.org/10.1371/journal.pone.0245590.t002

Results from the exploratory post-hoc analyses stratified by gender and age of the youth did not reveal any significant moderating effects for all models (see S2 File); hence we report results from the combined model. As recommended by Frazier and colleagues [53], moderation effects were not plotted for the results of the main hypothesis test in this study since they were not statistically significant. We further explored if parents’ gender did not suppress the effects of each parenting dimension, however, we still did not find statistically significant effects except for mothers’ demandingness which was found to significantly moderate the association between mothers’ sexual-risk communication and premarital sex, p = 0.04. Since these results were conducted as part of the post-hoc exploratory analyses, the simple slopes of the moderation effects are not presented here, but are included as S3 File.

Discussion

Contrary to propositions of the contextual model of parenting style, we found no statistically significant moderating effects of parenting style, nor did we find statistically significant associations of PYSRC or parenting style with premarital sexual debut. In the bivariate analysis, the finding that a higher proportion of males had engaged in premarital sex than females is consistent with the literature [8, 58]. Similarly, there was a high proportion of participants who had engaged in premarital sex among those whose peers had already engaged in premarital sexual activity (peer pressure), similar to other studies [6, 7] and earlier findings from Eswatini [16]. We also found that youth who had engaged in premarital sex were older than those who had not, consistent with findings by Waktole [5] and Young et al. [59]. This is not surprising since as youth grow older, they reach the age of sexual consent and therefore may feel more at ease initiating sexual activity.

Participants in this study reported lower parent-youth communication than in studies conducted in the United States [39, 40]. This could have been due to a majority of participants in our sample reporting largely having interactions with their mothers, and few having interactions with their fathers, and when summing up the scores, the overall PYSRC score was reduced. However, it is also possible that in reality, parents in Eswatini could be communicating less with their children about sexuality issues, since traditionally, this kind of communication is regarded as taboo, and instead, youth might be obtaining information about sexuality from elsewhere, such as from peers [16].

In the multivariate analysis, the absence of significant associations between PYSRC, or parenting style with premarital sexual debut suggests that other variables, which we did not measure in our study (confounders), might account for most of the variance in premarital sexual debut among youth in Eswatini, and that PYSRC and parenting styles are the least of them [31]. For example, Somers and Paulson [60] found that higher parental closeness in conjunction with child communication was unrelated to the sexual behavior of children, and instead, age was a stronger predictor. Therefore, in our case, it is also important to consider whether parenting style alone is adequate to explain child outcomes across populations and in different settings, or if it needs to be supplemented with alternative factors for some groups and not others [61]. For that reason, we cannot rule out that possibility in our study due to confounding. In fact, the borderline significant association between PYSRC and premarital sexual debut (see Table 2, model 3) suggests that some of the variance in premarital sexual debut is explained by a combination of PYSRC and parenting style, which suggests that PYSRC alone might be inadequate to explain the variance in premarital sexual debut, but when put together with authoritativeness, it explained some of the variance in premarital sexual debut.

However, we should point out that the direction of the borderline significant association (see text of multivariate results) is not what we had expected, as it showed that the greater the amount of PYSRC, the higher the odds of engaging in premarital sex. One possible explanation for this finding is that once parents notice that their children are dating or suspect that they are sexually active, they react by initiating or increasing discussions with their adolescents about sexual risks. Alternatively, it could also be that once the youth become sexually active, they approach their parents to seek clarity on certain topics related to their sexuality and therefore are the ones who initiate the discussions. However, the latter scenario is less likely in Eswatini as sexuality education is regarded as taboo, hence Swati children may be afraid to initiate conversations about sexuality topics with their parents. Unfortunately, due to the cross-sectional nature of the study, we could not establish the temporality between PYSRC and premarital sexual debut. The PTSRC-III also does not ask participants when sexual risk communication began.

We also did not find evidence of moderating effects of parenting style on the relationship between PYSRC and premarital sexual debut, despite the item mean score for the PSI-II being comparable with mean scores of previous studies [43, 45]. To the contrary, studies in other domains [62, 63] found significant moderating effects of parenting styles on child outcomes, while Schary and colleagues [64] found no such effects. One likely explanation for the absence of significant moderator effects in this study could be that perhaps parenting styles also act in other pathways other than those proposed by the contextual model of parenting style. For example, Li and colleagues [65] found that parenting styles mediate parent-child relationships. In light of these conflicting findings, there is a need for more studies using larger samples and longitudinal designs to further examine the mechanism through which these parental processes influence child outcomes.

It is also possible that the absence of significant main and moderation effects in this study could have been due to the cultural context of parenting styles and parent-child processes in Eswatini vis-à-vis other settings such as in Western societies, where much of the mainstream literature on parenting has been published. A recent qualitative study among adolescent in Tanzania found that most parents appeared to use an authoritarian style of parenting [66], suggesting cultural differences in parenting style in different context. Mainstream thinking and research about parent-child relationships have been dominated by Western cultural beliefs, which may have erroneously assumed that the meaning of parenting is similar across cultures [67]. However, such ideology may obscure important differences in what various cultures expect of and understand about parenting and parent-child relationships, especially because parenting practices vary from one culture to the next [68].

Phoenix and Husain [61] made strong arguments about bias in generalizing parenting styles across cultures and ethnicities. They argued that child outcomes may be affected by contextual or cultural factors, and that each culture has different goals and expectations of their citizens, such that, an effective parenting style elsewhere (e.g., authoritative parenting style in the United States of America) might not be as effective in other cultures [61]. This view is in line with an earlier study by Chao [69] who found that constructs of authoritative and authoritarian parenting were not relevant to Chinese-Americans. A recent systematic review of 428 studies on parenting styles and child outcomes also found that in western countries, associations of authoritarian parenting with academic achievement were less negative in Hispanic families than in non-Hispanic, White families. In all regions of the globe (with some regional variation), authoritative parenting was found to be associated with at least one positive child outcome, whereas authoritarian parenting was found to be associated with at least one negative outcome [70].

Parenting styles are also said to vary by geographical location; religious beliefs; parental culture; between and within individual parents; family and personal values; personality and temperament of the parents and child; socioeconomic status; and ethnicity [71]. Therefore, in this study, we could not rule out the possibility that global parenting styles might not account for premarital sexual debut of youth in Eswatini, and instead, other factors that interact with parental practices may be responsible for youth premarital sexual debut. For example, a previous study [72] found that subjective norms were the strongest predictors of intention for premarital sexual abstinence among in-school youth in Eswatini. This was also observed in other settings, where norms; values; social class; parental attitudes and approval; gender; and ethnicity were found to produce outcomes that ran counter to parenting style effects on child outcomes [61, 73]. Future studies should therefore investigate the influence of such factors on sexual behaviors of youth.

In exploratory post-hoc analyses, none of the youth-age or gender-stratified analyses produced any significant main or moderating effects. One might question whether sexual behavior in the mid-20’s is less related to parenting variables than sexual behaviors in teens, which might have contributed to the null result of the main hypothesis test in our study. We cross-tabulated the current age of the youth against their age at sexual debut to ascertain the proportion of adults who had their sexual debut in adulthood versus in childhood. Nevertheless, the cross-tabulation revealed that nearly 60% of older youth (18~24 years of age) had their sexual debut in adulthood. Having said that, we should point out the limitation in the measurement of the age-at-sexual-debut variable: participants were asked to choose from predetermined categories of 11 or younger, 12~16, and 17 or older, instead of 18 years or older. Unfortunately, that is how the variable is structured in the YSRBS which is the instrument we adopted for our study, and it was deemed appropriate at the time of data collection since the age of sexual consent in Eswatini was 16 years back then.

We further performed independent-sample t-tests to ascertain if parenting style and parent-youth sexual risk communication significantly differed by the current age of the youth and their age at sexual debut, and found no significant mean differences. In addition, since none of the youth-age or gender-stratified exploratory analyses showed statistically significant effects, it does not seem like the older youth in the sample suppressed the effect of parenting variables in the younger sample, which would have been the case had we found significant effects only in the younger sample (<18 years of age). In summary, as much as it is plausible that sexual behavior in older youth is less likely to be related to parenting variables than in younger youth, that does not seem to be the case in the current sample. One plausible reason could be that Swati parents generally do not talk to their children about sexuality issues (as discussed in detail above in the preceding text), independent of the stage or age of the child, such that the slope is constant across the adolescent and youth life course. Outputs from the cross-tabulations and t-tests are presented in the S4 File.

The only post-hoc analysis result of a significant moderating effect was found when parenting style dimensions were separately included in the models. The finding of simple slopes means that the effect of mother-youth sexual risk communication on premarital sex differed at different levels of demandingness, such that for low mothers’ demandingness scores (i.e., those below the mean), mother-youth sexual risk communication did not predict premarital sex, whereas it significantly positively predicted premarital sex at the mean level and at high mother demandingness (above the mean) scores (see S3 File). Again, the lack of information on the timing of communication between the mothers and their children limits our full interpretation of this finding as reverse causation is a possibility, as discussed above. More studies that include parent-child dyads are needed to further investigate this association separately based of the gender of the parents and youth.

Conclusions

This study is among the first to empirically test the contextual model of parenting style in a sub-Saharan African sample. Therefore, our findings contribute to the growing body of knowledge on parenting styles and child outcomes. We found no evidence of an association between PYSRC and premarital sexual initiation, and no evidence of moderating effects of parenting style between these two variables. In light of these findings and the study limitations, longitudinal studies are warranted to better understand the mechanisms through which parental practices and parenting styles influence youth sexual behaviors, especially in Africa.

Supporting information

Acknowledgments

We would like to extend our sincere gratitude to the students (and their parents) who agreed to participate in this study, and to the school managers for the administrative permission to enter their schools. We would also like thank Prof. Katherine M. Hutchinson and Prof. Nancy Darling for permission to use the PTSRC III and PS-II scales, respectively. We further thank Prof. Kuei-Ru Chou for her technical guidance at the conceptualization stage of the study, Dr. Yen-Kuang Lin for statistical guidance, and Lindelwa Portia Dlamini for her constructive comments on an early draft of the manuscript.