The associations between family characteristics and problematic Internet use among adolescents in Saudi Arabia

ABSTRACT

The current study assessed the family correlates of problematic Internet use (PIU) in a nationally representative sample of Saudi adolescents. Students (grades: 7–12; N = 2,546) from three cities in Saudi Arabia (random selection of schools) filled out a survey on Young’s Internet Addiction Test (YIAT) and family characteristics (e.g. parents’ socioeconomic status [SES], family harmony, family members’ Internet use, and parental Internet controls). A multilevel log-binomial regression assessed family correlates of PIU (YIAT ≥40). The mean (SD) age was 15.6 (±1.8) years; 54% were girls; PIU prevalence was 41.8%. A poor SES (OR = 1.2, 95% CI 1.0, 1.4), low family harmony (bottom quartile OR = 1.4, 95% CI 1.2, 1.7), no parental restriction/supervision (OR = 1.4, 95% CI 1.2, 1.8), and heavy Internet use by siblings (OR = 1.3, 95% CI 1.1, 1.5) were significantly associated with PIU. Certain identified family risk factors are amenable to interventions given the high PIU prevalence among Saudi adolescents.

KEYWORDS:

• Family characteristics
• Internet addiction disorder
• Saudi Arabia
• epidemiological factors
• adolescent

Introduction

Impaired control of one’s online activities is the key component of problematic Internet use (PIU) (Bleakley et al., 2016; Gómez et al., 2017). With PIU, a person chooses online over offline daily activities and has difficulty stopping or decreasing screen use in the face of negative consequences. PIU is a more appropriate term to characterize one’s impaired control over Internet use than Internet addiction (IA) in the absence of valid clinical diagnostic criteria for the latter (Griffiths et al., 2014). PIU has been classified in multiple ways; one is whether PIU refers to a particular type of Internet use (specific PIU of gaming, gambling, or social networking) or multidimensional Internet use (generalized). In addition, PIU has been studied in conjunction with several other mental health conditions (i.e. autism spectrum disorder, attention deficit hyperactivity disorder, and schizophrenia) since these conditions make a person more vulnerable to PIU (Kamolthip et al., 2022).

PIU among Saudi adolescents is gaining national attention. Its exact prevalence among this age group is not known because large-scale representative studies have yet to be conducted. Available data from around the world indicate that the global prevalence is between 4% and 25% (Nielsen et al., 2019; van den Eijnden et al., 2010; Wu et al., 2016). However, there are specific reasons to hypothesize that the PIU prevalence is even higher in Saudi Arabia. Saudi Arabia is a high-income country, and technology has been well-integrated into the society (98% internet penetration [World Bank, 2021]), which means nearly all adolescents have access to electronic devices. In addition, the hot weather limits outdoor activities during the day. High levels of PIU are of national concern because PIU is associated with physical and mental ill health and other consequences such as poor academic performance among adolescents (Chen et al., 2021; Nafee et al., 2018; Stavropoulos et al., 2021). Adolescents with PIU are more likely to report sleep problems, have difficulty maintaining a healthy weight, suffer from stress and depression, and have a poor quality of life (Alimoradi et al., 2019; Chen et al., 2021, 2020, 2021).

Current research proposes a variety of mechanisms for the development of PIU as well as addictions to the Internet and smartphones (Lin et al., 2023). Theorists have suggested that the personality trait of narcissism may be a mediating factor through which these behavioural addictions develop. People who are more narcissistic are more likely to engage online socially, get more enjoyment from online social contacts, and are more likely to self-promote on social media platforms (McCain & Campbell, 2018). Other studies suggest that lifestyle factors such as religiosity, age of initiation of Internet use, and screen time are associated with having PIU (Saquib et al., 2022).

Family plays a crucial role in problematic Internet use; PIU is more prevalent among families with single or divorced parents, interparental conflict, and parent-child conflict (Alimoradi et al., 2019; Nielsen et al., 2019). Similarly, less quality time with family members is associated with PIU among children (Li et al., 2014; Shek et al., 2019). Of the parenting styles, authoritative parenting by both parents was associated with the lowest PIU prevalence, while the combination of maternal authoritarian and paternal neglectful parenting style was associated with the highest prevalence (Lukavská et al., 2020). General characteristics such as parents’ employment and socioeconomic status, family functioning (i.e. harmony), and family members’ Internet use have been linked to PIU outside of Saudi society (Nielsen et al., 2019; Sayılı et al., 2021; Wartberg et al., 2014). However, several characteristics unique to Saudi society, such as a large family size and multiple wives, have not yet been assessed. A comprehensive understanding of family factors may lay the foundation for developing culturally sensitive and effective interventions.

Available Saudi studies on PIU are limited in number and quality. All studies collected samples locally, and therefore, did not represent the adolescent population at large. A few studies reported unadjusted/descriptive results only (Kolaib et al., 2020; Taha et al., 2019). The other studies that did perform multivariate modelling examined downstream correlates of PIU, such as sleep and quality of life, but did not focus on antecedent modifiable risk factors (Abdel-Salam et al., 2019; Barayan et al., 2018; Saquib et al., 2017). Additionally, those studies used regular logistic regression for outcomes that are common (>10%), which likely inflated the estimates for the identified factors (Diaz-Quijano, 2012).

Consequently, this multicity study was conducted among Saudi intermediate and high school students to estimate generalized PIU prevalence and to assess family-related risk factors. There were two a priori hypotheses: (a) the odds of having PIU would be higher for adolescents with no parental control or supervision over their Internet use compared with adolescents with both parental control and supervision, and (b) the odds of having PIU would be higher for adolescents with poor family harmony compared with adolescents with good family harmony.

Methodology

Study design

We conducted a cross-sectional study between January and March of 2020 among students enrolled in the governmental schools (grades 7–12) of major cities in Saudi Arabia. We planned to visit four cities (i.e. Buraidah, Dammam, Jeddah, and Riyadh), but we were not able to collect data from Dammam as the timing coincided with the national lockdown due to the COVID-19 pandemic. The Research and Development Office at the Ministry of Education (MOE) reviewed and approved the study protocol. Most Saudi schools (~83%) are under the jurisdiction of the MOE, and the government implements a nationally unified curriculum in Arabic.

Inclusion criteria

Both male and female students currently enrolled in grades 7 through 12 in a governmental school were included.

Sample size

We assumed a PIU prevalence anywhere between 20% and 50%. In 2016, there were ~ 1.8 million Saudi adolescents between 15 and 19 years (General Authority for Statistics Saudi Arabia, 2019). We needed to sample 3,456 students to estimate the prevalence of PIU (confidence level = 95%, margin of error = 2%, design effect = 4). We collected data from schools in Buraidah (n = 790), Jeddah (n = 979), and Riyadh (n = 810) (Dammam excluded); 33 records had missing data on the outcome, so the analytic sample was 2,546.

Sampling strategy

We used a stratified sampling strategy by level (i.e. intermediate vs. secondary) and by gender (male vs. female schools). The directorate of the MOE in each city supplied us with the list of schools. We selected 32 schools randomly from the list within each of the following strata: eight per city, four intermediate (two for boys and two for girls) and four secondary (two for boys and two for girls) schools. The research team contacted the school administration about enrolling in the study, and all selected schools agreed. This report includes data from 24 schools in three cities (excluding Dammam). Within each school, all students were eligible. School administrators selected classrooms from each grade level. The research team conducted the survey among all students inside those classrooms who gave consent.

Study procedure

Trained research assistants visited each school twice. On the first visit, they explained the study procedures to the students, invited them to participate, and gave them informed consent forms to be signed by their parents. On the second visit, the students completed the survey. Parental approval was assumed if the informed consent form was not returned (i.e. passive informed consent approach) (Eaton et al., 2004). It took the students 20 to 30 minutes to fill out the paper-based, self-administered questionnaire. Each survey questionnaire had a unique identification code, and fieldworkers returned them to the office of the project manager, who kept the surveys confidential under his supervision.

Pretesting

Prior to the fieldwork, we administered the questionnaire to five would-be eligible adolescents. We received feedback on the clarity of questions and response options, and we adjusted the language accordingly (Ruel et al., 2015).

Exposure variables (i.e. family factors)

Exposure variables included socioeconomic status (rich, upper-middle class, lower-middle class, or poor), parents’ marital status (married, divorced, or at least one parent deceased), number of father’s wives, number of siblings, and father’s and mother’s employment (yes/no).

Family harmony

We used the validated Arabic version of the 10-item short Scale of Family Atmosphere (SOFA) (Abdel-Khalek, 2016; Molloy & Pallant, 2002). Some example items were ‘I respect my parents,’ ‘I have a happy and close relationship with my mother/my father,’ and ‘My house is full of tension and disagreements.’ Each item was rated on a 5-point scale from strongly disagree to strongly agree. Scores were between 10 and 50, with a higher score indicating a more harmonious family atmosphere. For analysis, we categorized our participants by quartile value (Q1–Q4).

Family internet behaviour

We inquired from the participants about the Internet use (nonuser, light, moderate, or heavy) of their father, mother, brothers, and sisters; each family member’s use was dichotomized into heavy vs. all others. The participants also reported the control their parents exercised over their Internet use (no restriction or supervision, supervision only, restriction only, or both restriction and supervision).

Outcome variables

PIU was the primary and IA was the secondary outcome, both of which were determined with the validated Arabic version of Young’s Internet Addiction Test (YIAT) (Hawi, 2013). It is a 20-item scale that measures the characteristics and behaviours associated with a compulsive use of the Internet, such as escapism, dependency, etc (Young, 1998). Each item is rated on a 6-point Likert scale (not applicable, rarely, occasionally, frequently, often, always). We defined PIU as a summary score ≥ 40 and IA as ≥ 70, according to the literature (Guertler et al., 2014; Kitazawa et al., 2018).

Covariates

Covariables included city code (Buraidah, Jeddah, and Riyadh) and school code. Individual characteristics included age (≤13, 14, 15, 16, 17, ≥18 years), gender (male vs. female), nationality (Saudi vs. non-Saudi) and academic performance (grade point average [GPA]: excellent ≥ 90%, very good = 80–89%, good = 65–79%, fair = 50–64%, and poor < 50%).

Statistical analysis

We ran frequencies for all variables to identify implausible values and extreme outliers, which were checked against the hard copy file and were corrected. The quantity of missing data on the outcome was extremely low (n = 33, 1.3%); hence, we excluded them from the analysis. The covariate missing data ranged from 1%–4%; we substituted the missing values either with the mean (for continuous) or the most common response (for categorical) value. We reported frequencies of variables for the whole sample and between those who had PIU and those who did not; we compared them using the Chi-square test.

The original analytic plan was to use a hierarchical mixed model with the binary outcome (PIU vs. No PIU) to control for the potential nesting of data by city and school. However, these variables, when entered as random effects, were not significantly related to the outcomes. We also considered using an ordinary logistic regression, but it was not suitable for modelling the correlates of PIU because the prevalence of PIU was too common (estimate ~42%). Therefore, we used the log-binomial regression, which is less likely to overstate odds ratios (Barros & Hirakata, 2003; Diaz-Quijano, 2012). Diaz-Quijano showed how ordinary logistic regression significantly overestimates the effect, compared to log-binomial, when the outcome tends to be more common (Diaz-Quijano, 2012).

We ran an unadjusted model for each variable and considered it for the adjusted model if it had a p-value of 0.1 or less; parents’ education, number of fathers’ wives, and number of siblings did not qualify for the adjusted model. We adopted a block entry approach to model PIU (Grilli & Rampichini, 2007), which quantifies the contribution of each block separately and adjusts for the variables in the previous blocks. We entered geographical characteristics (i.e. city and school) in the first level, individual characteristics in the second level, and family factors in the third level. Each level contributed to a significant improvement of the model. We reported the odds ratio and 95% confidence interval for each variable. We ran a sensitivity analysis after excluding participants with a YIAT score ≥ 70 to determine whether risk factors were the same for moderate levels of impairment.

In addition to PIU, we also modelled IA (YIAT ≥70); we followed the same multilevel approach and included the same set of covariates. We analysed the data with SPSS version 24 and used a two-tailed test with an alpha of 0.05.

Results

Nonparticipation

A total of 3,000 informed consent forms were distributed (1,000 per city), and 2,579 students participated in the study (response rate 86%). The reasons for non-participation included absent students on the day of the survey, students’ refusal, and parental refusal (n = 421).

Sample characteristics

The greatest number of participants were from Jeddah. A slight majority of the sample were female (53.8%) and high school students (53.6%). More than two-thirds were Saudi nationals (69%), and half of the sample (49.8%) reported having an excellent GPA (Table 1).

Table 1. Demographic characteristics of a nationally representative study of problematic Internet use (PIU) among adolescents (grades 7–12) in Saudi Arabia (n = 2,546).

		Count	Overall	PIU=No	PIU=Yes	p
			%	%	%
City name	Buraidah	780	30.6	34.4	25.4	<.0001
	Jeddah	966	38.0	33.7	43.8
	Riyadh	800	31.4	31.9	30.8
School level	Intermediate	1,182	46.4	49.8	41.8	<.0001
	High school	1,364	53.6	50.2	58.2
Age (years)	≤13	373	14.7	17.3	11.0	<.0001
	14	345	13.6	14.5	12.2
	15	433	17.0	16.7	17.5
	16	497	19.5	19.2	20.0
	17	446	17.5	15.6	20.1
	≥18	452	17.8	16.8	19.1
Gender	Male	1,176	46.2	49.6	41.5	<.0001
	Female	1,370	53.8	50.4	58.5
Academic performance	Excellent	1,268	49.8	52.7	45.7	.006
	Very good	784	30.8	29.3	32.9
	Good	415	16.3	15.2	17.8
	Fair/Poor	79	3.1	2.8	3.6
Nationality	Saudi	1,756	69.0	71.3	65.8	.003
	Non-Saudi	790	31.0	28.7	34.2

An overwhelming majority of participants (83.7%) identified their parents’ social class as either rich or upper-middle class; 11.5% reported that either their parents were divorced or that one of them was deceased; 14% of participants’ fathers had multiple wives. Family size was generally large; 47.7% had between five and eight siblings, and 17.9% had more than eight siblings. One-fifth (21.6%) reported that their fathers were not employed, and three-fourths (73.4%) reported that their mothers were not employed (Table 2).

Table 2. Family characteristics of a nationally representative study of problematic Internet use (PIU) among adolescents (grades 7–12) in Saudi Arabia (n = 2,546).

		Count	Overall	PIU=No	PIU=Yes	p
			%	%	%
Parents’ marital status	Married	2,254	88.5	90.4	86.0	0.001
	Divorced	158	6.2	4.9	8.1
	At least one parent deceased	134	5.3	4.8	5.9
Number of father’s wives	1	2,121	86.0	87.5	83.9	0.014
	2	283	11.5	10.6	12.6
	≥3	62	2.5	1.9	3.4
Number of siblings	zero	26	1.1	1.2	0.9	0.60
	1–4	803	33.3	32.6	34.4
	5–8	1,149	47.7	48.6	46.5
	>8	430	17.9	17.6	18.2
Father’s employment	Working	1,922	78.4	79.7	76.7	0.08
	Not working	528	21.6	20.3	23.3
Mother’s employment	Working	642	26.6	26.1	27.2	0.57
	Not working	1,774	73.4	73.9	72.8
Socioeconomic status	Rich/UMC	2,132	83.7	86.6	79.8	<.0001
	LMC/Poor	414	16.3	13.4	20.2

UMC, upper-middle class; LMC, lower-middle class.

One-third (33.8%) reported that their parents did not supervise or place restrictions on their Internet use, whereas 16.3% reported that their parents both supervised and restricted their Internet use (Table 3). Most of the adolescents reported that their brothers (62%) and sisters (55%) were heavy Internet users, whereas only 20% reported that their parents were (Figure 1). The prevalence of IA and PIU in this sample were 3.7% and 41.8%, respectively.

Figure 1. Family member Internet use among a nationally representative study of problematic Internet use (PIU) among adolescents (grades 7–12) in Saudi Arabia (N = 2,546).

Table 3. Family environment and Internet-related behaviour of a nationally representative study of problematic Internet use (PIU) among adolescents (grades 7–12) in Saudi Arabia (n = 2,546).

	Count	Overall	PIU=No	PIU=Yes	p
		%	%	%
Parental control of Internet use
Full restriction plus supervision	416	16.3	19.3	12.2	<.0001
Restriction only	579	22.7	25.4	19.0
Supervision only	690	27.1	27.2	26.9
No restriction or supervision	861	33.8	28.1	41.9
Father’s Internet use
Not heavy user	2030	79.7	82.8	75.4	<.0001
Heavy user	516	20.3	17.2	24.6
Mother’s Internet use
Not heavy user	2045	80.3	83.0	76.6	<.0001
Heavy user	501	19.7	17.0	23.4
Brothers’ Internet use
Not heavy user	975	38.3	44.8	29.2	<.0001
Heavy user	1571	61.7	55.2	70.8
Sisters’ Internet use
Not heavy user	1158	45.5	53.0	35.0	<.0001
Heavy user	1388	54.5	47.0	65.0
Family harmony quartile
Q1 = 10–33	755	29.7	25.2	35.8	<.0001
Q2 = 34–35	784	30.8	30.9	30.7
Q3 = 36–37	488	19.2	20.5	17.3
Q4 = 38–50	519	20.4	23.4	16.2

Comparison of participants with and without PIU

In the unadjusted analysis, PIU among participants strongly varied by city, school level, age, and gender. The proportion of adolescents with PIU was highest in Jeddah; it was also higher among high school students and among female students. The proportion with PIU increased incrementally across the increasing age categories, except ≥18 years. The PIU prevalence was slightly higher among non-Saudi students and among those whose academic performance was less than excellent (Table 1).

PIU among participants significantly varied by their parents’ marital status and by their socioeconomic status. It was higher among those whose parents were either divorced or had one deceased parent. PIU was also higher among participants with a lower-middle class or poor background. PIU varied significantly by number of father’s wives but did not vary by parents’ employment or by number of siblings (Table 2).

PIU greatly varied by the participants’ family harmony, parental control of Internet use, and family members’ Internet use (p-values <.0001). For example, the proportion with PIU incrementally decreased across the increasing quartiles of family harmony (Q1 = 35.8%, Q2 = 30.7%, Q3 = 17.3%, Q4 = 16.2%). On the other hand, it incrementally increased with lesser parental control and/or supervision, from 12.2% with full restriction and supervision to 19.0% with restriction only, to 26.9% with supervision only, to 41.9% with no restriction or supervision (Table 3).

Adjusted family-related correlates of PIU

As a block, family factors explained a significant proportion of the variance in PIU (Chi-square = 104.4, df = 12, p < .0001) (Table 4). When adjusting for city, school, age, gender, nationality, and academic performance, the significant family correlates of PIU were socioeconomic status, parental control of Internet use, family Internet use, and family harmony. The odds of having PIU were 1.2 times higher for participants with a lower-middle class/poor background compared with participants with an upper-middle class/rich background (OR = 1.2, 95% CI = 1.02, 1.40, p = 0.03). Less parental control was associated with PIU; compared with adolescents whose parents exercised full control and supervision of their Internet use, the odds of having PIU were 1.3 times greater for participants whose parents provided supervision only (OR = 1.3, 95% CI = 1.04, 1.58, p =.02), and 1.4 times higher for those whose parents provided no supervision or restriction (OR = 1.4, 95% CI = 1.18, 1.76, p < .0001). Similarly, the odds of having PIU were 1.3 times higher among participants whose siblings also heavily used the Internet (brothers OR = 1.3, 95% CI 1.09, 1.45; sisters OR = 1.3, 95% CI 1.14, 1.51) and among participants who scored very low (i.e. Q1) in family harmony (OR = 1.4, 95% CI = 1.15, 1.65, p < .0001) (Table 5). In the sensitivity analysis, which excluded the small portion (3.7%) of adolescents with IA (YIAT ≥70), the overall findings remained the same with only small changes in the estimates (data not shown).

Table 4. Multilevel model building strategies for PIU among adolescents (grades 7–12) in Saudi Arabia (n = 2,546).

	Model 1	Model 2	Model 3
Block 1 (geographical factors):	χ ^2 = 34.9
City	df = 3
School	p < .0001
Block 2 (individual factors):
Age		χ ^2 = 21.6
Gender		df = 8
Nationality		p = .006
Academic performance
Block 3 (family factors):
Parents’ marital status
Socioeconomic status			χ ^2 = 104.4
Family Internet use (father, mother, brothers, sisters)			df = 12
Family harmony			p < .0001
Parental control of Internet use
Overall Model:
Chi-square (χ ^2)	34.9	55.4	159.6
Degrees of freedom (df)	3	11	23
p-value	<.0001	<.0001	<.0001

Table 5. Adjusted associations of PIU (YIAT ≥40) in a nationally representative study of adolescents (grades 7–12) in Saudi Arabia (n = 2,546).

Variables		OR	95% CI	p
Parents’ marital status
Married and together		Ref
Divorced/At least one parent deceased		1.2	0.98, 1.39	0.09
Socioeconomic status
Rich/UMC		Ref
LMC/Poor		1.2	1.02, 1.40	0.03
Parental control of Internet use
Full restriction plus supervision		Ref
Restriction only		1.1	0.86, 1.33	0.56
Supervision only		1.3	1.04, 1.58	0.02
No restriction or supervision		1.4	1.18, 1.76	<.0001
Father’s Internet use
Not heavy user		1.0
Heavy user		1.2	1.0, 1.34	0.06
Mother’s Internet use
Not heavy user		1.0
Heavy user		1.1	0.93, 1.26	0.31
Brothers’ Internet use
Not heavy user		1.0
Heavy user		1.3	1.09, 1.45	.002
Sisters’ Internet use
Not heavy user		1.0
Heavy user		1.3	1.14, 1.51	<.0001
Family harmony quartile
Q1 = 10–33		1.4	1.15, 1.65	0.001
Q2 = 34–35		1.2	1.01, 1.46	0.04
Q3 = 36–37		1.1	0.89, 1.36	0.37
Q4 = 38–50		Ref

YIAT, Young’s Internet Addiction Test; UMC, upper-middle class; LMC, lower-middle class The model is additionally adjusted for city, school, age, gender, nationality, and academic performance.

Academic performance for ‘not excellent’: OR = 1.2, 1.04, 1.34, p = .009 (Ref: excellent).

Discussion

There were several salient findings of this nationally representative study among adolescents in Saudi Arabia: (a) problematic Internet use (PIU) prevalence was substantial (~41.8%), but Internet addiction (IA) prevalence was relatively low (~3.7%); (b) PIU was significantly higher among girls than boys, among high school students than intermediate school students, and among non-Saudis than Saudis; and (c) several family-related factors such as poor socioeconomic status, low family harmony, no parental supervision alone or in combination with no restriction of Internet use, and heavy use among siblings were significantly associated with PIU after controlling for confounders.

The PIU prevalence greatly varies around the world (7.9%–55%) (Cam & Ustuner Top, 2020; Kumar et al., 2019; Laconi et al., 2018; Mazhari, 2012; Moreno et al., 2019), and its prevalence in Saudi Arabia (42%) stands at the upper end of that spectrum. There are several potential explanations for this high prevalence. The Internet penetration in Saudi Arabia has risen sharply in the last decade; it was 38% in 2009 and 96% in 2019 (World Bank, 2021). That means many more adolescents were connected to the Internet each year during this period with a resultant rise in PIU prevalence. A second potential explanation lies in the use of disparate assessment tools; some studies used the entire Young’s Internet Addiction Test, and others used only a portion or a modified version of it. Still others used completely different instruments, such as the Chen Internet Addiction Scale, the Compulsive Internet Use Scale, and the Mobile Phone Problem Use Scale (Dahl & Bergmark, 2020). A final explanation pertains to the harsh, hot weather and lack of extracurricular activities in Saudi Arabia. In many countries in Europe and Asia, there is a plethora of activities specific to adolescents, such as sports, summer camps, or cultural events. However, these activities are not widely available in the kingdom, so offline activity options are limited for adolescents. Studies suggest that PIU prevalence, along with other behavioural addictions, may have increased during the COVID-19 lockdown period in several countries (Alimoradi et al., 2022). This increase likely also occurred in Saudi Arabia because educational activities were conducted online for roughly 18 months, and there were restrictions on social and sports gatherings.

We found that relationships among family members greatly influence Internet use and the likelihood of PIU among adolescents. Although different studies have used various measures to assess family relationships, most studies show similar results, including the current study. The likelihood of adolescents having PIU is high in families with low harmony, i.e. high interparental conflict or low cohesion among members (Chang et al., 2015; Ko et al., 2015; Soh et al., 2018).

We did not find a similar study that assessed the degree to which family members use the Internet and whether that might be associated with the PIU among adolescents. Several studies showed that Internet addiction among adolescents decreased with an increased number of siblings or with better quality relationships among the siblings (Obeid et al., 2019; Öner & Arslantaş, 2018), while it increased with siblings using drugs and alcohol (Yen et al., 2007). There is an indication, however, that siblings, particularly the older ones, play an important role in shaping technology use among younger siblings, both in terms of content and time spent (Heinmäe & Siibak, 2016). These findings could be explained by the construct of ‘modelling’ (a.k.a., observational learning) from the Social Learning Theory (Bandura, 1986). Modelling suggests that learning takes place by watching others and repeating the observed behaviour without any direct reinforcement. Studies have shown that parents, caregivers, and possibly siblings can act as role models from whom adolescents learn behaviours such as Internet use (quantity and types) as well as a sedentary lifestyle (Lin et al., 2019; Vaala & Bleakley, 2015).

Parents address their children’s use of the Internet differently. Parents who do not supervise their children’s Internet use may not recognize that excessive Internet use is a potential problem, may be too busy with other areas of life to pay attention, or may be heavy Internet users themselves and do not feel it necessary to regulate their children’s Internet use (Lee & Kim, 2017; Wąsiński & Tomczyk, 2015). In contrast, there are parents who maintain strict supervision and/or restriction. The literature consistently shows that PIU is less common among adolescents whose parents monitor and regulate their Internet use (Clark, 2011; Martins et al., 2020). This may be related to socioeconomic status, but previous studies on socioeconomic status (i.e. parent income) and PIU had mixed results (Chang et al., 2015; C. S. T. Wu et al., 2016; J. Y. Wu et al., 2016). The current study supports the finding that adolescents in lower-income families were more likely to have PIU than those in high-income families. The socioeconomic status may have a combined effect with family harmony and parental controls over Internet use. Parents in low-income families could have longer working hours and need to dedicate more time to financially supporting the family, leaving less time to supervise their children’s Internet use.

Although the focus of the current study is PIU, one should note that PIU and IA are often assessed with the same instrument (e.g. Young’s Internet Addiction Test) but have different cut-off values: ≥ 40 for PIU and ≥ 70 for IA. Several Saudi studies of adolescents have varied widely in their reported IA prevalence. For example, a 2014 study reported a prevalence of 5.3% (Alhantoushi & Alabdullateef, 2014), and another in 2016 reported a 4.7% prevalence (Bafakih et al., 2016). On the other hand, a 2018 study found a prevalence of 41%, but it used a different instrument and did not provide the definition of IA (Khan & Gadhoum, 2018). The IA estimate of the current study (3.7%) is more in line with the earlier two studies.

There were notable strengths of this study: (a) a random selection of schools from multiple cities that included the full age distribution and adequate representation of female students; (b) use of validated instruments for exposure (i.e. family atmosphere, SOFA) and PIU outcome assessment (i.e. Young’s Internet Addiction Test) with a distinction between PIU and IA; (c) a comprehensive assessment of family factors that included the family structure, socioeconomic status, family member Internet use, and parental strategies towards adolescent Internet use; and (d) adoption of an appropriate modelling technique to avoid inflated associations between exposures and outcome.

One study limitation was that the data from one city were not collected because schools were closed during the pandemic. Another limitation was that the sample sizes from the included cities were roughly equal and not proportionate to their population sizes, which may have affected the representativeness of the sample. Family members’ Internet use was reported by the participants as a general characterization of each member and should not be interpreted as specific quantities. A caveat of this study is that the results are correlational and not causal due to the cross-sectional design. The PIU that we have reported is generalized PIU and cannot distinguish between the specific types of Internet use. Finally, participants with PIU may have had other mental health conditions that were not assessed.

The focus of this study was to assess the family factors related to PIU above and beyond individual factors. The models were adjusted for certain relevant individual factors; however, physical activity was not one of them. In this survey, there was no significant difference in physical activity between participants with PIU and those without PIU (published previously) (Saquib et al., 2022). However, other recent studies have shown that physical activity is positively associated with problematic social media use and problematic gaming (Huang et al., 2022; Liu et al., 2022; Saffari et al., 2022; Xu et al., 2022). It has been suggested that certain features in social media and games may prompt and motivate students to be active. More research is needed in this area.

Conclusions and recommendations

Saudi Arabia is a demographically young nation; 15% of its population is between the ages of 15 and 24 (General Authority for Statistics Saudi Arabia, 2019). Given that 42% of adolescents may have PIU, the absolute number of adolescents with this condition is very large. The findings of this study likely apply to adolescents across Saudi Arabia (i.e. generalizability), given the adequate random sampling from the major cities in the kingdom; the findings may also apply to other countries with similar cultures and family values.

A few of the significant family factors found in this study are modifiable, such as family harmony, parental supervision and control, and family members’ Internet use. Since family strongly influences the values and behaviour of Saudi adolescents, we recommend that prevention and intervention strategies target the whole family unit. Specifically, large-scale interventions and/or public health campaigns are needed to teach parents strategies for creating a more peaceful home environment, to monitor adolescents’ Internet use in order to guide them towards a healthy and balanced lifestyle, and to encourage all family members to use the Internet responsibly and not excessively. Future research should focus on programmes that teach children and adolescents about the concept of digital well-being (i.e. a healthy balance between online and offline activities). School administrators and parents could play an important role in mitigating the problem of PIU, by putting Internet use in perspective with other healthy and risky behaviours. They can inform students about the negative consequences of excessive Internet use and provide them methods to moderate their own use, such as digital well-being apps, self-notifications, and timers.

Author contributors

JS and NS: Design of the work, data acquisition, analysis, and drafting; AAS, MCC, ABL, AMA, ACB, AAM: Data interpretation and critical revision of manuscript. All authors have accepted responsibility for the entire content of this manuscript and approved its submission.

Acknowledgments

The authors wish to thank Dr. Fahad Abdullah Almatham, Prof. Muhsin Almuhsin, Dr. Syed Baquibillah, Dr. Ilias Hossain, Dr. Ahmed Rajab, and Dr. Saadi AlJundi for their help with the proposal development. We also thank the following people for their contributions—field supervisors for data collection: Abdullah AlKhani and Mohammed Ali (Jeddah), Saed Enabi and Amjad Chamsi Basha (Riyadh), Renad Aljarbue, Tawfiq Rajab and Azzam AlRobean (Qassim); data collection and data entry: Alhanouf Al-Rshide, Ghadir Al-Akhfash, Khadijah Al-Durrah, Maha Al-Enazi, Sara Al-Chalati, and Thekra Al-Khalaf; data cleaning and data management: Ayman Ibrahim; and English language editing: Erin Strotheide.

Disclosure statement

Dr. Colder Carras is the CEO of Gaming and Wellness Association, Inc., a non-profit dedicated to research and education about healthy video game play. She is also the CEO of the consulting firm Carras Colder Carras, LLC, and has conducted paid consulting for clients related to video game research and government grant making.

Data availability statement

Data will be made available upon reasonable request to corresponding author.

Additional information

Funding

This work was supported by a grant (Grant No. 4600000114) from the Research Development Office of the Ministry of Education, Saudi Arabia. It did not have any role in the design, data collection, analysis, or manuscript preparation.