Challenging the stigma of premarital sex in urban Afghanistan

Abstract

Premarital sex is normatively unacceptable in Afghanistan, yet rapid social and cultural transformation in the country may be changing these traditional norms. In dialogue with cross-national scholarship, we examine attitudes toward premarital sex and experience of premarital sexual behaviours among urban Afghan youth. We use data from 1256 never married individuals aged 15–29 surveyed in ten Afghan cities. The results show that respondents, on average, have moderately liberal attitudes toward premarital sex and that almost one quarter of them had engaged in either sexual foreplay or penetrative intercourse. Multivariable results indicate that premarital sexual attitudes and behaviours were significantly associated with several individual-level characteristics, family and intergenerational relationships, and social ties and interactions. At the same time, instructive gender variations emerge. The findings illustrate the dynamics of premarital sexual relations in this and similar contexts where such relations are traditionally stigmatised but multifaceted, societal changes increasingly challenge this stigma.

Introduction

Catalysed by rapid social and cultural changes, premarital sexual activity among young people has been increasing worldwide in recent decades (Shahid, Abu Bakar, and Wahab 2017; Lat et al. 2020). However, it is still common in cultural discourses on sex to separate ‘good sex/normal sexuality’ from ‘bad sex/abnormal sexuality’ (Rubin 1984). These cultural discourses delimit a very small portion of human sexual capacity as sanctifiable, safe, healthy, mature, legal and politically correct (Rubin 1984). Thus, culture lies at the heart of how individuals, families, communities and societies define and express sexuality (Tabahi 2020). Universally, religion provides a major cultural context for, and determinant of, these definitions and expressions. With regards to premarital sex, in particular, it has been argued that Muslims are less likely to report having had premarital sex than adherents of other major religions (Adamczyk and Hayes 2012), since in many Muslim societies social and religious values maintain a strong taboo over premarital sex and stigmatise individuals, particularly women, who become involved in it (O’Donnell, Utomo, and McDonald 2020). In such societies, marriage remains the prerequisite for sexual intimacy and the key mechanism for both procreation and social reproduction; accordingly, premarital sex is morally and socially unacceptable. This normative stance is clearly gendered: the preservation of girls’ virginity provides not only an index of family honour but is also a gender-biased cultural practice (Ghanim 2015).

This study is focused on one such society, Afghanistan, where family honour – and consequently family status – are maintained through the institution of marriage, with its key role in controlling women’s and men’s sexuality (Smith 2009). In Afghanistan, pre/extramarital sex is considered a zina offence: if either party is married, the offence is classified as adultery; if both are unmarried, the offence is fornication. Traditionally, zina entails severe punishment – adultery is punished by stoning and fornication by flogging – even though under the formal legal system a more typical punishment for either type of zina is imprisonment (Mir Hosseini 2011; EASO 2017).

However, similar to other predominantly Muslim contexts, recent social and political changes in Afghanistan, such as rapid urbanisation, Westernisation, the spread of new media and communication technologies, migration, and accompanying demographic shifts have generated new aspirations and expectations for intimate and family life (Sadeghi and Bakhtiari 2018; Wimpelmann et al. 2019). As part of these structural and ideational transformations, even very conservative societies have experienced a rise in premarital sex (Jayakody, Thornton, and Axinn 2007; Majumdar 2018). In this study, we use recent survey data to examine attitudes toward and experience of premarital sex among unmarried Afghan youth by relating these attitudes and behaviours to individual traits, family characteristics and relationships, and peer influences.

Background and hypotheses

To explain changes in sexual attitudes and behaviours, two main theoretical perspectives have been proposed: a structural perspective and an ideational change perspective. The former emphasises the role of modernisation and changing location within the social structure, especially within gender and family systems, in shaping sexual attitudes and behaviours (Brooks and Bolzendahl 2004; Wellings et al. 2006; Lesogorol 2008). The ideational change perspective, in contrast, assumes that changes in attitudes toward sex and in sexual behaviour are part of a broader process of ideational transformation (Brooks and Bolzendahl 2004). From this perspective, sexual attitudes and behaviours are shaped by changing cultural values and norms which favour the social acceptance of novel sexual practices (Hatfield, Luckhurst, and Rapson 2010). The importance of such ideational factors in explaining family and other demographic changes is well established (Inglehart and Norris 2003; Lesthaeghe 2014; Sadeghi and Agadjanian 2019). There is a growing recognition, however, that structural and ideational forces act in synergy with one another to bring about change in sexual attitudes and practices (Jayakody, Thornton, and Axinn 2007). Here, we focus on several key dimensions that represent the fusion of the two perspectives.

Sexuality and gender equality are inextricably linked, and both have been highly politicised issues in many Middle Eastern settings ever since modernisation reforms were initiated in the nineteenth century (Ilkkaracan 2008). However, in recent decades, the region has seen a considerable transformation of sexual discourse while struggling to fit sexuality, reproductive rights and women’s rights, in general, into an emerging gender-egalitarian narrative (Tabahi 2020). While political cultures have proved slow to change in the wake of the Arab Spring, growing freedom of expression, widening debate over human rights and personal liberty, and increasing civic activism have opening space for direct, even if discreet, challenges to sexual taboos across the region (El Feki 2015). However, the majority of the population in the Middle East, as elsewhere in the Muslim world, still believes that engaging in sex before marriage is not acceptable (Statista Research Department 2014). Yet, due to persistent taboos regarding sexuality in Muslim societies, investigation of premarital sex in these societies is relatively rare, and it is unclear how theories and evidence originating from outside the Islamic world apply to Muslim contexts. This study contributes to filling this important gap.

Synthesising the theoretical and empirical evidence from cross-national scholarship and applying it to the Afghan context, we examine three groups of factors that might shape attitudes toward and practice of premarital sex among Afghan youth – socio-demographic factors, family environment, and social relationships. The first socio-demographic characteristic that we consider is gender. Previous studies have generally shown small gender differences in overall sexual attitudes and behaviours. Petersen and Hyde (2011) have argued that gender differences in attitudes toward premarital sex have decreased since the early 1990s. However, studies still find that men tend to have more premarital sexual experience than women (Cubbins and Tanfer 2000; Seth et al. 2012). Yet, research by Cubbins and Tanfer (2000) showed considerable gender differences in sexual behaviours in the USA due to differences in men’s and women’s family roles, employment, religious behaviours and past sexual experience. Seth et al. (2012) examined gender differences in sexual behaviours in the USA and found that female adolescents were more likely than male adolescents to refuse sex or to avoid a situation that might lead to sex; they were also less likely to have a casual sexual partner than their male counterparts. Based on this evidence, we expect to find no significant gender differences in attitudes toward premarital sex (Hypothesis 1a), yet we hypothesise that men are more likely to have premarital sex than women (Hypothesis 1 b).

Previous studies have shown a strong association between education and sexual attitudes and premarital sexual behaviours. For example, Zuo et al. (2012) found in three Asian urban settings that young people with university education had more liberal premarital sexual attitudes than their less educated one counterparts. However, Yip et al. (2013) in their study of unmarried youth in Hong Kong found that young adults with university education were nearly half as likely to engage in premarital sex as those with lower educational levels. Similarly, Berliana et al. (2018) concluded that less educated people are more likely to have premarital sex in Indonesia. Our second pair of hypotheses reflects this scholarship: namely, educational attainment will be positively associated with liberal sexual attitudes (H2a) and negatively associated with engaging in premarital sex (H2b).

Religiosity is an important cultural factor shaping sexual attitudes and premarital sex. A negative relationship between religiosity and premarital sexual attitudes and behaviours has been documented in several studies. A review by Rostosky et al. (2004) showed that religiosity is associated with a delay the sexual debut, even though the association is stronger among female adolescents than males. Farmer, Trapnell, and Meston (2009) found in the USA that religion-based spirituality, fundamentalism and paranormal beliefs were associated with a lower likelihood of engaging in sexual behaviour. Earle et al. (2007) in the study of never married college students in the USA found that religiosity and involvement in sexual activity were inversely related for both women and men. Gyimah et al. (2013) reported that greater religiosity was associated with more conservative views about premarital sex in urban Kenya. Motamedi et al. (2016) showed that being less religious or being secular were important determinants of liberal sexual attitudes in Iran. We, therefore, hypothesise that religiosity will be negatively associated with liberal sexual attitudes (H3a) and premarital sex (H3b).

Beyond religiosity, other aspects of cultural traditionalism typically associated with ethnic identity may affect sexual attitudes and behaviours. For example, in their study of sexual attitudes among US college students, Ahrold and Meston (2010) reported that Asian Americans held more conservative attitudes than did their Hispanic and Euro-American peers. In their study, for both Hispanics and Asian Americans, greater acculturation also predicted sexual attitudes similar to those of Euro-Americans. Afghanistan is a multi-ethnic country, with Pashtuns, the largest ethnic group, being more culturally conservative compared to other groups (Barfield 2011). We therefore hypothesise that Pashtuns will be less likely to have liberal sexual attitudes (H4a) and engage in premarital sex than members of other ethnic groups (H4b).

The family environment constitutes the second group of factors. Prior research has documented significant family socioeconomic status (SES) differentials in premarital sexual attitudes and behaviours. Thus, Wong and Lam (2013) found that among Chinese adolescents in Hong Kong family income was positively associated with acceptance of premarital sex. Li et al. (2009) reported a low level of family income as the most important predictor of having had sexual intercourse in last three months among adolescent migrant workers in Shanghai, China. Wepukhulu et al. (2012) found that poverty and economic deprivation were positively linked to premarital sexual activity among adolescent girls in Kenya. In sum, these studies suggest a positive SES gradient for acceptance of premarital sex but a negative SES gradient for engagement in premarital sex. Accordingly, we hypothesise that liberal sexual attitudes will be positively associated with family SES (H5a) while experience of premarital sex will be negatively associated with it (H5b).

Studies have also examined young people’s relationships with their parents as a predictor sexual attitudes and behaviours. Frantz, Sixaba, and Smith(2015) reported that co-residence with parents was associated with delaying or reducing engagement in risky sexual behaviours in several African contexts. Podhisita, Xenos, and Varangrat (2001) found in Thailand that good parent-adolescent communication may reduce risk-related behaviours among adolescents. Using longitudinal data from a survey of teenagers in Scotland, Wight, Williamson, and Henderson (2006) showed that low parental monitoring predicts early sexual activity for both sexes. Gyan Bahadur and Basel (2013) found in Kathmandu, Nepal, that young people who had good relationship with their parents were less likely to engage in premarital sex than were young people who had poor relationship with their parents. Based on this evidence, we expect both liberal sexual attitudes and premarital sex to be negatively associated with the strength of family and intergenerational ties, as proxied by co-residence with parents (H6a & H6b), good quality of parent-children relationship (H7a & H7b), and talking about sexual matters with parents (H8a & H8b).

Our last group of factors includes social ties and interactions. It is well established that peer group influence can greatly increase the likelihood of having liberal sexual attitudes and of engaging in sexual behaviours. For example, Sieving et al. (2006) in the USA, found that, regardless of other factors, the proportion of a young person’s friends who were sexually experienced was positively associated with the odds of that person’s own sexual debut. Chiao and Yi (2011) demonstrated in Taiwan that peer pressure as measured by perceptions of friends’ sexual behaviour was strongly associated with engagement in premarital sex among adolescents. Bingenheimer, Asante, and Ahiadeke (2015) showed that perceived peer views favouring sex increased the odds of sexual debut in Ghana. Therefore, we hypothesise that liberal sexual attitudes and premarital sex will be positively associated with sexual experience of close friends (H9a & H9b).

Social media is an important informative and communicative source of sexual experiences. The effects of exposure to social media on sexual attitudes and behaviours are also well documented. Thus, the findings by Adegboyega (2019) demonstrated that social media are often used among youth in Kwara State, Nigeria, to exchange erotic messages and watch pornographic films and videos conducive to such sexual behaviours as masturbation. Lin, Liu, and Yi (2020) showed that sexually explicit media exposure in adolescence predicted early sexual debut, unsafe sex and multiple sexual partnerships. Accordingly, we hypothesise that exposure to social media will be associated with liberal sexual attitudes and greater likelihood of engaging in premarital sex (H10a & H10b).

In addition to testing these hypotheses, we examine variations in the effects expected under Hypotheses 2 to 10 by gender. The evidence reviewed above, however limited, does not point to any clear direction in most such variations. This part of our analysis is therefore exploratory.

Data and methods

Sample

We analysed data from The Study of Sexual Attitudes and Behaviors of Youth in Afghanistan. As part of this study, a survey, directed by the first author, was carried out in 2017 by the Afghan Institute for Strategic Studies (AISS). The survey used multistage cluster sampling of never married young people aged 15–34 residing in ten Afghan cities of varied size. The sample size was determined using a single population proportion formula with the following assumptions: confidence level of 99%, margin of error of 4%, a design effect of 2, and non-response rate of 15%. The target sample size was 1,479; 1,350 individuals were successfully interviewed, yielding a response rate of 91.3%.

In each city, five neighbourhoods were randomly selected, then households were randomly selected in each sampled neighbourhood, and finally, one never married resident aged 15–34 was chosen for interview in each selected household. Male and female interviewers interviewed male and female respondents, respectively, in the Dari or Pashto languages. Most of the interviews were conducted at the door of the respondents’ houses, but some respondents preferred to be interviewed in a public place such as parks and coffee shops. Here, we limit our analytic sample to respondents aged 15–29 (N = 1256) because in this context of relatively early marriage, individuals who remain unmarried into their thirties constitute a very atypical segment of society.

The survey, preceded by qualitative exploration aimed at designing and refining the survey questions, collected information on attitudes toward marriage and sexuality, sexual experiences, education, socioeconomic status, religiosity, health and various other characteristics.

Measures

Our analysis had two outcomes: 1) liberal attitudes toward premarital sex; and 2) premarital sexual experience. Because having had premarital sexual experience is highly correlated with attitudes toward premarital sex, the analysis of liberal attitudes toward premarital sex was limited to respondents who had never engaged in sexual activity. Liberal attitudes toward premarital sex were measured by an 8-item scale developed based on the qualitative explorations preceding the survey (Sadeghi and Bakhtiari 2018). Respondents were asked to rate the following eight items using a 5-point Likert-type scale ranging from 0 (strongly disagree) to 4 (strongly agree): 1) Premarital sexual intercourse is a basic instinct need and has to be fulfilled; 2) It’s natural for unmarried young people to have sexual relations with the opposite sex; 3) Sex education should be given to unmarried youth; 4) The virginity of a girl is the key precondition of marrying her; 5) Premarital sex reduces marital satisfaction in future life; 6) Premarital sex is a great guilt and should be avoided; 7) If a sexual relationship between an unmarried man and a woman before marriage is revealed, they have to marry each other; (8) Any pre-marital sexual relationship is considered a crime and must be punished by stoning. We reverse recoded the answers to the last five items. The scale has Cronbach’s alpha α = 0.76 and ranges from 0 (very conservative) to 32 (very liberal).

Experience of premarital sexual activity was defined as having ever engaged in sexual activity with a person of the opposite sex (the survey did not include questions on same-sex relations). Respondents were first asked the following question: ‘Have you ever had any sexual experience or contact with the opposite sex?’ Respondents who answered yes, were then asked: ‘What kind of sexual experience have you had?’ Response options included: sexual talking, hugging, petting, kissing, oral sex, vaginal sex, and anal sex. We distinguished between what we labelled as sexual foreplay (sexual talking, hugging, petting, kissing, and oral sex) and penetrative sexual intercourse (vaginal or anal sex). Accordingly, this outcome was operationalised as a three-level ordinal scale: none; only foreplay; and intercourse.

The predictors were of three categories. Individual socio-demographic characteristics included gender (women, men) and education (no education, lower secondary, higher secondary, bachelor’s degree, and master’s/doctoral degree). Religiosity was assessed on the basis of responses to the following three questions: 1) How often do you participate in religious activities and practices; 2) How much do you believe that God governs all our actions and behaviours; and 3) To what extent do you consider your-self a religious person? Based on the combined index score (α = 0.68), respondents were classified into three categories: very religious, moderately religious, and mildly religious. The last individual predictor was ethnicity (Pashtun vs. other ethnic groups such as Hazara, Tajik, Uzbek, etc.).

The second category of predictors was family characteristics and relationships. This included perceived family SES (poor, medium, rich), co-residence with parents (living with parents, not living with parents), perceived quality of relationship with parents (cordial, neither cordial nor conflictual, conflictual), and frequency of talking about sexual matters with parents (never, sometimes, often). The third category, social ties and interactions, included two predictors: presumed/known sexual experience of close friends (no close friends have had such experience; at least one close friend has had such experience; does not know/not sure if any friend has had such experience), and social media exposure (none vs. at least some).

In addition to these predictors, the multivariable models controlled for respondent’s age group (15–19, 20–24, and 25–29 years) and city of residence (Kabul or other cities). The distribution of covariates is shown in Table 1.

Table 1.

Characteristics of survey respondents.

Variables		N	Percent
Individual characteristics
Gender	Woman	621	49.4
	Man	635	50.6
Education	No education	179	14.3
	Lower secondary	338	26.9
	Higher secondary	221	17.6
	Bachelor’s degree	457	36.4
	Master’s/doctoral degree	61	4.9
Religiosity	Very religious	729	58.0
	Moderately religious	366	29.1
	Mildly religious	161	12.8
Ethnicity	Pashtun	363	28.9
	Other	893	71.1
Family and intergenerational ties
Perceived family SES	Poor	284	22.6
	Medium	728	58.0
	Rich	244	19.4
Co-residence with parents	Living with parents	900	71.7
	Not living with parents	356	28.3
Quality of relationship with parents	Cordial	611	48.6
	Neither cordial nor conflictual	546	43.5
	Conflictual	99	7.9
Talking about sexual matters with parents	Never	704	56.1
	Sometimes	442	35.2
	Often	110	8.8
Social ties and interactions
Presumed/known sexual	No close friends had sexual experience	456	36.3
experience of close friends	At least one close friend had sexual experience	440	35.0
	Does not know/not sure	360	28.7
Social media exposure	No exposure	376	29.9
	At least some exposure	880	70.1
Controls
Age group	15–19	344	27.4
(Average = 22.1)	20–24	566	45.1
	25–29	346	27.5
City of residence	Kabul	558	44.4
	Other city	698	55.6
Total		1256	100.0

Analysis

We began with a bivariate analysis of the relationships of the predictors with premarital sexual attitudes and behaviour. Then, we fitted OLS and ordinal logit regression models to test our hypotheses on attitudes toward premarital sex and premarital sexual experience, respectively. For each outcome, we fitted an all-gender model and separate models for men and women. All the multivariable models included the same set of predictors and controls.

Results

Descriptive results

Descriptive results for liberal attitudes toward premarital sex among respondents who had not sexual intercourse are presented in Table 2. The mean score of attitudes towards premarital sex was 14.2. As can be seen, these attitudes had no statistically significant association with gender, perceived family SES, and talking about sexual matters with parents. The other predictors showed significant associations: liberal attitudes were positively related to education, sexual experience of close friends, and social media exposure, but negatively related to belonging to the Pashtun ethnic group, religiosity, co-residing with parents, and the quality of relationship with parents.

Table 2.

Liberal attitude toward premarital sex among respondents who had no sexual intercourse, by predictor variables (means and SDs).

		Liberal attitude toward premarital sex (scale 0–32)
Variables		M	SD	N
Individual socio-demographic characteristics Gender	Woman	13.9	6.2	593
	Man	14.4	7.0	541
Education***	No education	13.2	5.6	158
	Lower secondary	13.4	6.2	313
	Higher secondary	14.7	6.8	200
	Bachelor’s degree	14.5	6.8	408
	Master’s/doctoral degree	17.8	7.8	55
Religiosity***	Very religious	12.8	5.9	661
	Moderately religious	14.9	6.4	336
	Mildly religious	18.6	7.9	137
Ethnicity***	Pashtun	13.2	6.2	336
	Other	14.6	6.7	798
Family and intergenerational relations
Perceived family SES	Poor	13.7	6.3	251
	Medium	14.2	6.7	661
	Rich	14.6	6.7	222
Co-residence	Living with parents	13.5	6.1	826
with parents***	Not living with parents	15.9	7.5	308
Quality of	Cordial	13.8	6.2	556
relationship	Neither cordial nor conflictual	14.3	6.8	496
with parents**	Conflictual	16.3	7.3	82
Talking about sexual	Never	13.8	6.8	619
matters with parents	Sometimes	14.4	6.3	409
	Often	14.7	6.4	106
Social ties and interactions
Presumed/known	No close friends had sexual experience	11.8	5.4	448
sexual experience of At least one close friend had sexual experience		17.5	7.3	345
close friends***	Does not know/not sure	13.9	5.8	341
Social	No exposure	12.9	6.1	352
media exposure***	At least some exposure	14.8	6.7	782
Total		14.2	6.6	1134

Notes: t- test and F-test are used; significance

^*p < .05;

^**p < .01;

^***p < 001.

The descriptive results for the correlates of premarital sexual experience are presented in Table 3. At the time of the survey, 68% of men and 86% of women had never had any sexual experience, 14% (18% of men and 10% of women) had had only sexual foreplay, and 10% (15% of men and 5% of women) had engaged in sexual intercourse. Table 3 reveals statistically significant associations between premarital sexual experience and all individual-level factors, except education, as well as with family characteristics and intergenerational relations, and with social ties and interactions. Specifically, having engaged in premarital sexual activity was negatively associated with religiosity, belonging to the Pashtun ethnic group, family socioeconomic status, co-residing with parents, good parent-children relationship, and talking about sexual matters with parents. Conversely, the sexual experience of close friends and social media exposure were positively associated with having engaged in premarital sexual activity.

Table 3.

Experience of premarital sexual activity, by predictor variables (%).

		Ever had premarital sex
Variables		None	Only foreplay	Intercourse	N
Individual characteristics
Gender***	Woman	85.5	10.0	4.5	621
	Man	67.7	17.5	14.8	635
Education	No education	73.7	14.5	11.8	179
	Lower secondary	82.0	10.6	7.4	338
	Higher secondary	79.2	11.3	9.5	221
	Bachelor’s degree	72.9	16.4	10.7	457
	Master’s/doctoral degree	72.1	18.0	9.9	61
Religiosity***	Very religious	79.1	11.5	9.4	729
	Moderately religious	77.6	14.2	8.2	366
	Mildly religious	62.1	23.0	14.9	161
Ethnicity***	Pashtun	84.3	8.3	7.4	363
	Other	73.2	16.0	10.7	893
Family and intergenerational ties Perceived family SES*	Poor	75.2	13.2	11.6	284
	Medium	76.5	14.3	9.2	728
	Rich	75.4	16.7	8.0	244
Co-residence with parents**	Living with parents	79.8	12.0	8.2	900
	Not living with parents	68.3	18.3	13.5	356
Quality of relationship	Cordial	77.1	13.9	9.0	611
with parents*	Neither cordial nor conflictual	76.7	14.1	9.2	546
	Conflictual	71.7	11.1	17.2	99
Talking about sexual matters	Never	71.9	16.0	12.1	704
with parents**	Sometimes	82.4	10.2	7.4	442
	Often	82.7	13.6	3.7	110
Social ties and interactions
Presumed/known sexual experience of close friends***	No close friends had sexual experience	95.0	3.2	1.8	456
	At least one close friend had sexual experience	51.6	26.8	21.6	440
	Does not know/not sure	83.6	11.1	5.3	360
Social media exposure**	No exposure	84.0	9.6	6.4	376
	At least some exposure	73.3	15.6	11.1	880
Total		76.5	13.8	9.7	1256

Notes: Chi-square test is used; significance

^*p < .05;

^**p < .01;

^***p < .001.

Multivariable results

The results of the multivariable all-gender and gender-specific models are displayed in Table 4. Section 4.A.1 shows the results of the OLS model predicting liberal sexual attitudes for the entire sample. There was a small and non-significant gender difference in liberal attitudes toward premarital sex (H1a supported). Liberal attitudes toward premarital sex were positively associated with university education, confirming H2a, but this positive association was significant only for men (4.A.2 and 4.A.3). Supporting H3a, liberal attitudes toward premarital sex were negatively associated with religiosity. Net of other factors, the score of liberal sexual attitudes was lower among Pashtuns, compared to respondents from the other ethnic groups, but in gender-specific models this difference was statistically significant for men only (H4a partially supported). Family SES was not significantly related to liberal sexual attitudes, offering no support to Hypothesis 5a. Liberal sexual attitudes were negatively associated with co-residing with parents (H6a) and with good parent-child relationships (H7a). However, the gender-specific models show that H6a was supported only for women, while H7a was supported only for men. Talking about sexual matters with parents (H8) was not significantly associated with having liberal sexual attitudes, even though women who sometimes talked about sexual matters with parents were more likely to express liberal sexual attitudes, compared to women who did so often. As we hypothesised, liberal sexual attitudes were positively associated with sexual experience of close friends (H9a) for both genders. Finally, exposure to social media (H10a) also showed a positive association with liberal sexual attitudes in the all-gender model. However, this association was significant only in the women’s model. These variables, together with age and city of residence, explained 23% of the variance in attitudes toward premarital sex in the combined model, but 31.6% in the men’s model and 19.5% in the women’s model.

Table 4.

Multivariable models predicting liberal attitudes toward premarital sex (OLS) and experience of premarital sexual activity (ordinal logit), by gender.

		4. A. Liberal attitude toward premarital sex scale Beta (OLS)			4. B. Experience of premarital sex Odds ratio (Ordinal Logit)
		1. All	2. Men	3. Women	1. All	2. Men	3. Women
Individual characteristics
Gender	Woman®
	Man	0.02			2.28***
Education	No education®
	Lower secondary	0.04	0.04	0.03	0.65	1.01	0.41
	Higher secondary	0.06	0.09*	0.02	0.73	0.88	0.52
	Bachelor’s degree	0.08*	0.12**	0.04	0.75	0.99	0.43*
	Master’s/doctoral degree	0.11**	0.19***	0.06	0.49**	0.54*	0.32**
Religiosity	Very religious®
	Moderately religious	0.12***	0.13***	0.10**	1.01	1.04	0.99
	Mildly religious	0.22***	0.30***	0.14***	1.36*	1.24	1.47*
Ethnicity	Pashtun®
	Other	0.07*	0.11 **	0.03	1.97***	2.76***	1.02
Family and intergenerational ties
Perceived family SES	Poor®
	Medium	0.01	0.01	0.02	1.06	1.32	0.77
	Rich	0.03	0.04	0.05	1.27	1.73*	0.64*
Co-residence with parents	Living with parents®
	Not living with parents	0.08*	0.03	0.11**	1.03	0.98	1.10
Quality of relationship with parents	Cordial®
	Neither cordial nor conflictual	0.01	0.03	0.01	0.97	1.12	0.87
	Conflictual	0.08**	0.13***	0.04	1.33*	2.08*	0.83
Talking about sexual matters with parents Sometimes	Never®	0.06	0.03	0.10*	0.71 **	1.02	0.46**
	Often	0.03	0.06	0.01	0.62*	1.01	0.57**
Social ties and interactions
Presumed/known	No close friends had
sexual experience of close friends	sexual experience®
	At least one close friend had sexual experience	0.35***	0.33***	0.34***	14.78***	14.94***	17.33***
	Does not know/not sure	0.14***	0.08*	0 17***	3.46***	2.69**	4.86***
Social media exposure	No exposure At least some exposure®	0.07*	0.04	0.09*	1.23	0.88	1.47*
Control variables
Age group	15–19®
	20–24	−0.05	−0.05	−0.04	0.83	1.29	0.45*
	25–29	0.02	0.02	0.02	1.17	2.11**	0.69
City of residence	Other city® Kabul	0.05	0.06	0.02	1.95***	1.62**	3.87***
R2 for OLS, Pseudo R2 for Logit		22.9	31.6	19.5	33.7	34.8	32.3
Model fit (F-test for OLS, Chi-square for Logit)		14.95***	11.64***	6.51***	344.2***	198.2***	132.7***
Number of cases		1134	541	593	1256	635	621

Notes: Premarital sex attitude model excludes respondents who ever had sex; ® reference category; significance

^*p < .05;

^**p < .01;

^***p < .001.

Section 4.B shows the results of the ordinal logit model predicting sexual experience. As we hypothesised, the likelihood of having engaged in premarital sexual activity was higher among men than women (H1b). In the all-gender model (4.B.1), engaging in premarital sexual activity was negatively associated with education only for masters/doctoral degree, compared to those with no education, even though education had a stronger effect on women’s (4.B.2) than on men’s (4.B.3) likelihood of premarital sexual activity (H2b partially supported). Religiosity was negatively associated with the likelihood of premarital sexual activity, but this negative association was significant only in the women’s model, thus partially supporting H3b. As predicted by H4b, the likelihood of having engaged in premarital sex activity was lower among Pashtuns than among other ethnic groups, but the disaggregation of the sample by gender shows that this association was present only among men.

The results of the all-gender model did not confirm H5b that higher family SES was associated with lower likelihood of premarital sexual activity. However, the gender-specific models showed that being in the rich tier, relative to the poor tier, was positively associated with premarital sexual experience among men but negatively among women. Co-residence with parents showed no net association with premarital sexual experience (H6b not supported). In line with our hypotheses, in the all-gender model, good parent-child relationship, relative to conflictual relationship, was a significant positive predictor of the likelihood of having engaged in premarital sexual activity (H7b), whereas talking about sexual matters with parents (H8b) showed a significant negative association with it. However, the effect of parent-child relationship was significant only for men, whereas the effect of talking about sexual matters with parents was significant only for women.

With respect to the third theoretical dimension of interest – social ties and interactions, in support of our hypothesis, having close friends with known/perceived sexual experience showed a very strong positive association with having engaged in premarital sexual activity (H9b). In comparison, time spent on social media was not significantly related to experience of premarital sexual activity in the all-gender model (H10b), even though the predicted positive association was found for women. Overall, the variables included in the all-gender model explained 34% of the variance in the likelihood of premarital sexual experience, 35% in the men’s model and 32% in the women’s model.

Discussion and conclusion

In this study, guided by fused structural and ideational change theoretical perspectives, we investigated premarital sexual attitudes and behaviours of Afghan urban youth. Our results point to instructive variations in premarital sexual attitudes and behaviours across and within the three dimensions of theoretical interest - individual socio-demographics, family background and relationships, and social ties and interactions.

With respect to the first dimension, our analyses supported our hypotheses predicting a small/insignificant gender differences in premarital sexual attitudes yet higher likelihood of engaging in premarital sexual activity among men compared to women. These findings resonate with those of other studies that reported gender similarity in sexual attitudes (e.g. Petersen and Hyde 2011) and those that found greater experience of premarital sex among men (e.g. Cubbins and Tanfer 2000; Seth et al. 2012). They further illustrate how nominally gender uniform normative expectations may diverge from actual gendered practices. Female virginity is highly valued in conservative Afghan society, as it is seen as a symbol of modesty and family pride. Girls who are deemed to have had sex before marriage face public shame, and even death. Girls also often face invasive and degrading forced virginity tests to determine whether their hymen is intact (Pazhohish 2017). Thus, premarital sex for Afghan women is treated as particularly ‘bad, abnormal, unnatural, and damned sex’, using Rubin’s (1984) terminology, and therefore poses much greater social risks for unmarried women than for unmarried men.

Our finding that educational attainment was positively associated with liberal sexual attitudes aligns with previous studies (e.g. Hager 1996; Zuo et al. 2012) showing that more educated young people are more likely to have more liberal sexual attitudes. Education may reflect exposure to modernising ideas that increase acceptance of premarital sexuality (Thornton and Camburn 1987). Yet, in our sample, that association was clearly present only among men. At the same time, we found that education is negatively associated with engaging in premarital sexual activity, especially among women. This finding conforms to those of several studies that have shown that more educated young people are less likely to engage in premarital sex (e.g. Yip et al. 2013; Berliana et al. 2018).

The negative association of religiosity with liberal premarital sexual attitudes and premarital sex detected in our analyses, parallels findings of other studies (Rostosky et al. 2004; Farmer, Trapnell, and Meston 2009; Gyimah et al. 2013; Motamedi et al. 2016) and illustrates the importance of religiosity as an ideological and cultural force shaping sexual attitudes and behaviours (cf. Inglehart and Norris 2003; Adamczyk and Hayes 2012). Yet, notably, the relationship of religiosity with sexual behaviour was not statistically significant in the men-only model.

Our findings also support the hypothesis that members of a more conservative ethnic group, the Pashtun, tend to have less liberal sexual attitudes and are less likely to have had premarital sexual activity than members of other ethnic groups. These findings add to the cross-national evidence on ethnic variations in sexual attitudes and behaviour (e.g. Simon and Paxton 2004; Ahrold and Meston 2010; Bacchus 2017). However, this association was statistically significant only for men in our sample.

For our second theoretical focus – family context and relationships – our all-gender analyses did not support the hypothesis on the positive association of the family SES with liberal sexual attitudes and the negative association with premarital sexual behaviours, although the lack of such associations was also reported in the literature (e.g. Wong 2012). However, the gender-specific models showed an intriguing divergence in the statistical effects of SES on premarital sexual experience – a positive effect for men but a negative effect for women. It may be that given the importance of daughters’ premarital virginity to family reputation, parents in higher SES families exercise greater control/supervision over girls to ensure their virginity. In comparison, for men, higher SES may offer greater opportunities for sexual encounters.

As we expected, overall, co-residing with parents and good parent-child relationship was associated with less liberal sexual attitudes. However, the significant association of liberal sexual attitudes with co-residing with parents was confirmed only for women and with good parent-child relationship was supported only for men. In addition, we found no overall association between sexual attitudes and talking about sexual matters with parents, even though women who occasionally talked about these matters were more likely to have liberal sexual attitudes than women who did not talk about them at all. It should be noted that the majority of respondents said that they never talked with parents about sexual matters. The findings for premarital sexual experience indicate that good parent-child relationship matters only for men and talking with parents about sex only for women, in decreasing the likelihood of having engaged in premarital sexual activities. While generally conforming to previous studies (Podhisita, Xenos, and Varangrat 2001; Wight, Williamson, and Henderson 2006; Guilamo-Ramos et al. 2012; Gyan Bahadur and Basel 2013) showing a negative association of the strength of family and intergenerational ties with premarital sex, our findings add potentially important gendered nuances to these relationships. Also, the lack of any net association between co-residence with parents and the likelihood of premarital sexual activity, suggests that co-residence with parents does not necessarily prevent unmarried youth from engaging in premarital sex.

For our last theoretical focus, we found that sexual experience of peers was positively and very strongly associated with both having liberal sexual attitudes and the likelihood of having engaged in sexual activity. Similar patterns have been reported in previous studies (Small and Kerns 1993; Sieving et al. 2006; Chiao and Yi 2011; Bingenheimer, Asante, and Ahiadeke 2015). As Libby, Gray, and White (1978) noted, friends’ sexual behaviour exerts a powerful influence legitimising and normalising the desirability and acceptability of premarital sex. Finally, our findings indicated that exposure to social media had distinctly gendered effects: it was positively associated with both liberal sexual attitudes towards and engagement in premarital sexual activity among women but not among men. These findings suggest an important qualification to the scholarship on the connection between social media use and sexual behaviour social media exposure in this context at least (e.g. Adegboyega 2019; Lin, Liu, and Yi 2020). It is important for future research to examine the gendered exposure to social media’s sexual contents in the Afghanistan context.

Limitations

We acknowledge that some of the detected gendered patterns do not lend themselves to straightforward explanations. Social desirability bias in self-reported sexual attitudes and behaviours may also have affected the results. Also, this study did not enquire about same-sex sexual relations. Finally, our findings, drawn from an urban sample and cannot be generalised to entire Afghanistan.

Conclusion

This study constitutes an important step toward a better understanding of the nature of premarital sexual relations in this understudied rapidly changing context, where such relations are traditionally stigmatised but multifaceted societal changes increasingly challenge that stigma. Future studies, using more diverse and nationally representative samples, will help to fully capture variations in premarital sexual attitudes and behaviours in Afghan society.