Introduction
One of the primary leisure activities for people today is using digital media, with the two dominant digital media activities being gaming and social media use. Screen time has increased steadily during the past decade, and recent statistics show that adolescents and young adults now spend between 7 and 8 hours/day using screen media, with over 4 hours on gaming and social media (Kemp,
2022; Rideout et al.,
2022). For most people, their digital media use is enjoyable and unproblematic, but there are concerns that an increasing number of people are developing addiction-like use (Meng et al.,
2022). In the latest Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association [APA],
2013), Internet Gaming Disorder (IGD) was presented as a diagnosis in need of further study. Given the potential of social media use to have negative psychosocial consequences, it has been suggested that Social Media Disorder (SMD) should also be considered as a diagnosis (van den Eijnden et al.,
2016). A recent study indicated that the prevalence of SMD in adults is equal to or higher than IGD (e.g., Burén et al.,
2021), and that the symptoms of both IGD and SMD are related to a range of negative outcomes, such as depression, anxiety, low self-esteem, and poor social relations (e.g., Pontes,
2017; Teng et al.,
2021). Several previous instruments have been developed to measure problematic gaming or social media use. For gaming, a recent meta-analysis identified 32 measures (King et al.,
2020). However, many of these instruments have limitations, and none include both IGD and SMD in the same measure, The present study therefore aimed to introduce a new rating instrument encompassing both gaming and social media, although providing separate measures for these two types of digital media addictions.
Defining IGD and SMD
According to the DSM-5 (APA,
2013), IGD is defined using the following nine symptoms, out of which at least five symptoms should be present to meet the criteria for IGD: 1) preoccupation with gaming; 2) withdrawal symptoms when not allowed to game; 3) need to increase time spent gaming (i.e., tolerance); 4) Inability to reduce playing/unsuccessful attempts to quit gaming; 5) loss of interest/hobbies due to gaming; 6) continued excessive use of gaming despite psychological problems; 7) lying about the amount of time spent gaming; 8) using gaming to relieve negative moods; 9) jeopardizing relationships or career/educational opportunities due to gaming.
In a relatively recent meta-analysis that included 53 studies and over 225,000 participants from 17 countries, the pooled prevalence rate for IGD was 1.96% after adjusting for sample biases (Stevens et al.,
2021). Several studies have also found IGD to be significantly associated with a range of comorbid conditions, including depression, anxiety, low self-esteem, poor social relations, and sleep problems (Cheng et al.,
2018; Lam,
2014; Müller et al.,
2015; Ostinelli et al.,
2021; Teng et al.,
2021).
When introducing IGD in the DSM-5, criticism was raised that other forms of compulsive digital media activities, such as social media use, were excluded (e.g., Kuss et al.,
2017; Müller,
2017). One argument was that addictive use of games and social media is part of the same overarching construct of digital media addiction, with addictive gaming and social media users sharing underlying risk factors and comorbid conditions (Andreassen et al.,
2016; Hussain & Griffiths,
2018; Wartberg et al.,
2020; van den Eijnden et al.,
2016). It has also been argued that the nine symptom criteria that are presented in the DSM-5 for IGD are applicable also for SMD if just replacing the word “gaming” with “social media use” (e.g., van den Eijnden et al.,
2016). Using these or similar criteria, a relatively recent meta-analysis (Cheng et al.,
2021) that included 63 samples with over 34,000 participants from 32 countries found that the pooled prevalence for SMD was 5%.
Regarding the co-occurrence of IGD and SMD, several studies have found a significant association between IGD and SMD symptom severity, but correlations have been small to medium in size (Andreassen et al.,
2016; Pontes,
2017; van den Eijnden et al.,
2018, but see Reer et al.,
2021 for an exception). In addition, the overlap between individuals meeting the full symptom criteria for IGD or SMD is often relatively small (Wartberg et al.,
2020). It has therefore been argued that social media addiction and gaming addiction should be regarded as separate but interrelated constructs (c.f., Andreassen et al.,
2016; Griffiths,
2018).
Previous Measures of IGD and SMD Symptom Severity
Several instruments assessing gaming have been presented, but many of them have important limitations. First of all, as shown in a relatively recent review (King et al.,
2020), most instruments were created prior to the introduction of IGD in the DSM-5. Only eight measures included all nine symptom criteria. Second, several available scales use a yes/no response format, which may increase the risk of overestimating the prevalence of IGD (Pontes & Griffiths,
2015). Several reviews and meta-analyses (Fam,
2018; Paulus et al.,
2018; Stevens et al.,
2021) have shown large variations in the prevalence of IGD, with some studies reporting rates as high as 50% (Hur,
2006), which is unreasonably high for non-clinical samples. In addition, it has been argued (APA,
2013) that viewing psychiatric symptoms as varying along a dimensional rather than only as a discrete category is the most sensible approach, and this further emphasizes the need to not use a yes/no response format when assessing gaming.
Fewer instruments have been developed to assess SMD. The two most commonly used instruments are the six-symptom Bergen Social Media Addiction Scale (Andreassen et al.,
2016) and the nine-symptom Social Media Disorder Scale (van den Eijnden et al.,
2016). There is, to our knowledge, no instrument available that measures addictive use of both gaming and social media, which should be considered an important limitation. It has been argued that such an instrument should preferably measure IGD and SMD symptom severity using identical items as this would allow for a more direct comparison between these two types of digital media addiction (Reer et al.,
2021; van Rooij et al.,
2017). Most available scales only assess IGD, which is more common among males than females (Stevens et al.,
2021). Thus, excluding assessments of social media addiction increases the risk of missing problematic use of digital media in females. Including both types of digital media use in the same measure would also allow researchers to assess the respective activities’ differential associations with mental health problems. This may be important because previous research has suggested that the underlying deficits of IGD and SMD may be at least partially different. For example, it has been found that emotion dysregulation is more strongly related to problematic use of social media compared to gaming (Leménager et al.,
2016). This finding was replicated in a more recent study investigating a broader number of neuropsychological deficits (i.e., executive deficits, delay aversion, and emotion dysregulation), although only for women (Soares et al.,
2023). However, another study failed to find a significant group difference between internet gaming addicts and social network addicts with regard to emotional inhibitory control (Dieter,
2017).
IGD and SMD as Uni- or Multidimensional Constructs
IGD is today described as a unidimensional construct in the DSM-5 (APA,
2013). However, before IGD was formally introduced, it was suggested that it is important to distinguish between symptom criteria related to heavy involvement in gaming (e.g., preoccupation and tolerance), which are common in the population, and symptom criteria related to negative consequences (e.g., displacement, and loss of interest), which may be less common but more indicative of problematic use (Charlton & Danforth,
2007). A few studies have found support for these two factors for IGD (Brunborg et al.,
2015; Wichstrøm et al.,
2019), but most studies have argued for a one-factor solution (e.g., Lemmens et al.,
2015; Pontes & Griffiths,
2015; Poon et al.,
2021). However, it has been stated (e.g., Billieux et al.,
2019) that a limitation of previous studies is that they have only used factor analysis to confirm a one-factor structure without exploring alternative factor structures. Concerning SMD, a cross-cultural study using the nine-symptom Social Media Disorder Scale found support for a one-factor model in 34 out of 44 countries (Boer et al.,
2022). In the other 10 countries, findings were inconsistent. In summary, there is a need for more studies examining several alternative factor structures for both IGD and SMD.
Sex and Age Differences
As mentioned above, males are more likely to develop IGD than females, with a recent meta-analysis showing a prevalence rate of 6.5% for males and 2.5% for females (Stevens et al.,
2021). These sex differences have been argued to be explained by worse inhibitory control, higher craving for games, and higher reward sensitivity in males compared to females (Dong & Potenza,
2022). Regarding SMD, a Dutch study including over 6600 adolescents found that the prevalence rate of SMD was higher for females (4.1%) than males (2.9%; Boer et al.,
2021). This is in line with a previous meta-analysis (not reporting prevalence rates), which found a significantly larger effect size for females compared to males (Su et al.,
2020). However, another meta-analysis failed to find significant sex differences (Cheng et al.,
2021). It has been speculated that females may be more at risk of SMD than males because they more often use social media to fill social needs and have a stronger inclination to invest in close social relationships (Andreassen et al.,
2017; Krasnova et al.,
2017).
Although many available scales assessing digital media addiction have been used for both adolescents and adults, the appropriateness of these scales for different age groups has not been assessed. This should be regarded as a limitation, as differences between adolescents and adults could influence their use of digital media and its potential negative consequences. Previous research has, for example, shown that compared to adults, adolescents’ social life and digital media use are more intertwined; they more often use digital media to build their identity (Allen et al.,
2014; Davis,
2013), they are more sensitive to peer pressure (Brown & Anistranski,
2020), and they are more inclined to take risks online (Prencipe et al.,
2011). With regard to prevalence rates, a recent meta-analysis (Gao et al.,
2022) found that IGD was slightly more common among young adults (10.4%) than among adolescents (8.8%). However, for social media addiction, a meta-analysis (Cheng et al.,
2021) found that adolescent samples had a higher prevalence rate (35%) than both university students (23%) and adults (19%). The very high prevalence rates in these two meta-analyses are most likely a result of the included studies using many different scales, and many studies did not operationalize addiction according to the DSM-5 criteria (APA,
2013).
Aims of the Present Study
To address the limitations of previous instruments assessing digital media addiction, the present study introduced a new brief screening instrument referred to as the Gaming and Social Media Questionnaire (GSMQ-9), which includes the nine symptom criteria for both IGD and SMD. As described above, addictive use of gaming and social media should be regarded as separate but interrelated constructs. The GSMQ-9 was therefore created to independently assess IGD and SMD symptom severity within the same rating instrument. More specifically, we aimed to examine the following issues:
1.
The reliability and factor structure of the GSMQ-9 among adolescents and adults
2.
Associations between IGD and SMD symptom severity, as measured by the GSMQ-9, and psychosocial outcomes (i.e., psychosomatic problems, self-esteem, social relations, and quality of life).
3.
The proportion of the participants meeting the symptom criteria for IGD and SMD based on self-ratings on the GSMQ-9 and differences between a) adolescents versus adults, b) males versus females.
Based on theoretical formulations (Charlton & Danforth,
2007) and previous empirical research examining several alternative factor solutions (e.g., Brunborg et al.,
2015; Wichstrøm et al.,
2019), we hypothesized that a two-factor solution, one related to heavy involvement and one related to negative consequences, would provide the best fit to our data. Second, we hypothesized that higher IGD and SMD symptom severity would be associated with higher psychosomatic and social problems, as well as lower self-concept and lower quality of life (e.g., Pontes,
2017; Teng et al.,
2021). Third, we hypothesized that males would have higher rates of IGD than females (Stevens et al.,
2021), whereas females would have higher rates of SMD than males (e.g., Boer et al.,
2021; Su et al.,
2020). Finally, regarding age differences, too few studies have examined differences in prevalence rates between adolescents and university students to allow for any a priori hypotheses.
Study 1: Measure Development
The Gaming and Social Media Questionnaire (GSMQ-9) aimed to be a self-report instrument for adolescents and adults, measuring addictive use of gaming and social media. The main idea was to create a short screening instrument with identical items for both two types of digital media addiction, although allowing the participants to provide separate ratings for gaming and social media. Responses were made on a 5-point Likert scale ranging from 0 ("Strongly disagree") to 4 (Strongly agree"), with higher scores indicating higher symptom severity. The aim was to derive one item for each one of the nine criteria for IGD as presented in the DSM-5 (APA,
2013). For each item, the participants were considered to meet a symptom criterion if they obtained a score of ≥ 3.0 on the scale ranging from 0 to 4. In accordance with the DSM-5 (APA,
2013), five out of nine symptoms were required to meet the full criteria for IGD or SMD.
When developing the instrument, we first generated 30 items related to the nine DSM-5 criteria for IGD (APA,
2013). These were then discussed with clinical psychologists (
n = 5), three of whom were also researchers. All of them had more than 10 years of previous experience working with patients with behavioral addictions. We also conducted interviews with young adults and parents (
n = 12). During these interviews, each one of the 30 items were presented one at a time and the persons being interviewed were asked to provide comments regarding the relevance and clarity of each item. The list of 30 items were presented and the persons being interviewed were asked to judge the clarity and relevance of each item. In addition, a small survey (
n = 22) was conducted with university students to determine how long the survey took to complete. These students were also instructed to provide written feedback if they found any of the questions difficult to answer or irrelevant. Finally, we selected the nine items that 1) best captured the content of the nine criteria for IGD as they are presented in the DSM-5 (APA,
2013); 2) were unambiguous (i.e., easy for the participants to interpret; not having multiple meanings), and 3) had observed scores that covered the full range of possible values on the 5-point Likert scale. The selection of the final items was based on discussion and consensus decision within the research group (the final version of the questionnaire is presented in
Supplementary Material). The items for the criteria “loss of interest” and “jeopardizing career/relationships” were different for the adult- and the adolescent versions to ensure the age appropriateness of the instrument (e.g., we asked about jeopardizing work/studies for young adults, whereas we asked about jeopardizing relations with family members for adolescents).
The GSMQ-9 was originally made in Swedish and Italian, which are the native languages of the authors. There was no significant difference between the Italian and the Swedish participants with regard to SMD symptom severity,
t(993) = 0.18,
p = .86). However, for IGD symptom severity, the Swedish sample exhibited a higher mean score compared to the Italian sample,
t(692.40) = 2.11,
p = .035, although the effect size was small,
d = 0.14). The final version of the GSMQ-9 is presented in Table
2 (English translation), and
Supplementary Material.
Discussion
The present study indicated that the GSMQ-9 appears to be a reliable screening instrument for assessing IGD and SMD in adolescents and adults. For both age groups, two factors were identified: Heavy Involvement and Negative Consequences. The proportion of participants meeting symptom criteria for IGD was 1.4% for adults and 1.8% for adolescents, whereas the proportion of participants meeting the criteria for SMD was 4.0% for adults and 2.6% for adolescents. For adults, the two factors for IGD and SMD were positively related to all three psychosocial outcomes. Significantly stronger positive associations were found for Negative Consequences compared to Heavy Involvement. For adolescents, higher levels of Negative Consequences for gaming, but not Heavy Involvement, were significantly associated with poorer quality of life. For social media, higher levels of both Heavy Involvement and Negative Consequences were significantly associated with poorer quality of life. Similarly to what was found for adults, associations were significantly stronger for Negative Consequences compared to Heavy Involvement.
In line with our hypothesis, the results of the present study provided strong initial support for a two-factor solution, one related to heavy involvement and one related to negative consequences, for both IGD and SMD. However, more studies with different samples are needed to further validate the factor structure of the GSMQ-9. Although our two-factor solution is in line with findings from a few previous studies (Brunborg et al.,
2015; Charlton & Danforth,
2007; Wichstrøm et al.,
2019), most previous research (e.g., Boer et al.,
2021; Lemmens et al.,
2015; Poon et al.,
2021; Paulus et al.,
2018; van den Eijnden et al.,
2016) has argued that IGD and SMD should be regarded as a unidimensional construct. The reason for this view appears to be largely based on the description of IGD in the DSM-5 (APA,
2013), which has led many researchers to only use confirmatory approaches rather than comparing alternative models (Billieux et al.,
2019). Based on both our own findings and previous theoretical formulations (Charlton & Danforth,
2007), we argue that there are some advantages to viewing IGD and SMD as involving two dimensions – one related to heavy involvement and one related to negative consequences. First, taking both these separate but overlapping phenomena into consideration when screening for IGD and SMD could possibly lower the risk of over-diagnosing these two types of digital media addictions. For example, previous research (Burén et al.,
2021; Rehbein et al.,
2015; Wichstrøm et al.,
2019) has shown that symptoms of heavy involvement are common among individuals with IGD, but also among those not meeting the symptom criteria, whereas symptoms of negative consequences are less common but highly predictive of digital media addictions once present. Second, as both the present study and previous research (Deleuze et al.,
2018) have shown that symptoms related to negative consequences are more strongly related to negative psychosocial outcomes, individuals with these symptoms may have a greater need for support than those who only have symptoms related to heavy involvement. Third, it could be argued that our two-factor solution is in line with the DSM-5 (APA,
2013) as both symptoms and negative consequences are required to meet the diagnostic criteria for many psychiatric disorders (e.g., Attention Deficit Hyperactivity Disorder). It has even been proposed, and our results could be taken to be in line with this proposition, that symptoms related to heavy involvement and negative consequences should be regarded as two separate domains within DSM-5 and that symptoms within both domains should be required for a diagnosis (cf. Wichstrøm et al.,
2019). Fourth, research conducted during the COVID-19 pandemic has shown an increase in the use of screens but not necessarily an increase in the negative consequences associated with social media use (e.g., Cauberghe et al.,
2021; Pandya & Lodha,
2021). This further emphasizes the need to distinguish between heavy usage and negative consequences and shows that, at least during some circumstances, a high screen time may even provide benefits.
If acknowledging that IGD and SMD can be divided into two factors, a related issue concerns to what extent this needs to be considered when screening and diagnosing IGD and SMD. This issue has been highly debated, with some researchers arguing that heavy involvement in digital media use is not a good indication of addiction (Billieux et al.,
2019; Charlton & Danforth,
2007). In a Delphi study including 29 international experts from 21 countries addressing diagnostic validity, clinical utility, and prognostic value for IGD, the symptom criteria related to heavy involvement were viewed as being less important for the disorder compared to those related to negative consequences (Castro-Calvo et al.,
2021). The results of the present study cannot solve this debate but emphasize the need to regard symptoms related to heavy involvement and negative consequences as separate but related constructs. An important avenue for future research would be longitudinal studies investigating the relative importance of these two factors for identifying individuals at risk of developing IGD or SMD.
The criteria loading on the factor Heavy Involvement found in the present study are similar to those identified in previous studies (e.g., Deleuze et al.,
2018; Wichstrøm et al.,
2019). However, in the present study, escape loaded on the factor Negative Consequences, whereas two previous Norwegian studies investigating IGD (Brunborg et al.,
2015; Wichstrøm et al.,
2019) found that it loaded on Heavy Involvement. We do not find this inconsistency across studies very surprising, as one can easily consider that this symptom criterion is related to both factors, or that it constitutes a factor of its own. In the DSM-5 (APA,
2013), the symptom criterion escape is defined as using digital media to relieve negative feelings. As such, escape may function as a form of “self-medication” for psychological distress (Király et al.,
2015). Others have suggested that digital media may function as a tool for satisfying basic psychological needs that are impossible to satisfy in real life, such as having many friends or being competent (Bender et al.,
2020). Thus, using digital media as an escape may not lead to negative consequences, but rather vice versa (i.e., that negative consequences in life lead to digital media use). Similarly, escape is not a direct example of heavy involvement in digital media, but an individual who is using digital media as an escape is likely to have larger incentives to have heavy involvement in this activity compared to those who only use digital media because they offer pleasurable activities. It has previously been argued that escapism motives are common in humans and may, therefore, not indicate disordered use of digital media (Király et al.,
2015). Escape has also been questioned for its clinical relevance to IGD due to mixed views concerning its clinical and diagnostic utility, as well as prognostic value (Castro-Calvo et al.,
2021).
The sample sizes in the present study were too small to determine the prevalence of IGD and SMD. However, we still considered it important to investigate what proportion of the participants met the symptom criteria for IGD and SMD as a further test of our new instrument. The proportion meeting the symptom criteria ranged between 1.4% and 1.8% for IGD and between 2.6% and 4.0% for SMD, figures that are comparable to those presented in relatively recent meta-analyses (Cheng et al.,
2021; Stevens et al.,
2021). This indicates that the GSMQ-9 most likely does not over- or underestimate prevalence rates for IGD or SMD, whereas high heterogeneity in prevalence rates (i.e., 0.16-50%) have been reported for IGD in previous studies using other rating instruments (Fam,
2018; Paulus et al.,
2018; Stevens et al.,
2021).
In line with our hypothesis and previous research, the proportion of males meeting the symptom criteria for IGD was higher than females (e.g., Stevens et al.,
2021). However, we did not find significant sex differences for SMD. Previous studies examining sex differences have shown mixed results, with one meta-analysis failing to find significant sex differences (Cheng et al.,
2021), whereas other studies have shown that social media addiction is more prevalent among females compared to males (e.g., Boer et al.,
2021; Su et al.,
2020). One explanation for these inconsistencies in findings might be differences in how social media addiction was operationalized, with sex differences often being found when examining SMD symptom severity, whereas sex differences are not found for SMD diagnosis (e.g., Burén et al.
2021). It is also possible that sex differences in social media use have decreased over time as more and more gamers are using social media to discuss gaming experiences, which might have led to increased use of social media among males. Further research using larger sample sizes is needed to explore sex differences for SMD diagnosis.
The present study did not find any differences in prevalence rates between adolescents and adults for SMD or IGD. As few previous studies have examined age differences, we did not form an a priori hypothesis regarding this issue. However, there are indications that the prevalence of IGD is higher among adolescents compared to adults (Gao et al.,
2022), whereas the prevalence of SMD is higher for adults compared to adolescents (Cheng et al.,
2021). Differences in findings could perhaps be explained by the fact that we included university students (primarily 20–30 years old), whereas previous studies have used a larger age range. Differences could also result from the small number of participants meeting the diagnostic cut-off, especially for IGD, which limited our statistical power to detect differences between adolescents and adults.
Finally, in line with our hypothesis, higher IGD and SMD symptom severity were associated with higher levels of psychosomatic and social problems, and lower levels of self-concept and quality of life. It is interesting to note that SMD symptom severity was also shown to be equally or even more strongly related to the negative psychosocial outcomes compared to IGD symptom severity. Although it is important to take into consideration that correlations do not imply causation, previous experimental research suggests that social media has an immediate causal effect on emotional distress (e.g., Lee et al.,
2020). Benefits from reducing use or taking a break from social media on mental health measures have also been reported (Allcot et al.,
2020; Hunt et al.,
2018; Tromholt,
2016). We, therefore, believe that it is important to question why IGD, but not SMD, has been recognized within the DSM-5.
Limitations and Directions for Further Studies
Some limitations of the study need to be acknowledged. First, the study used convenience samples which may limit the generalizability of our findings. Second, the use of self-report measures might have introduced biases and led to an overestimation of the associations between digital media and psychosocial outcomes. Third, although the present study found prevalence rates that were similar to those presented in other studies, further studies are needed to determine the best cut-off when using items measured on a Likert scale. Fourth, as we wanted to minimize responder fatigue, we chose not to include other established measures of IGD or SMD. Further research investigating the validity of the GSMQ-9 is therefore needed. Fifth, regarding the last DSM-5 criterion for IGD (i.e., “jeopardizing relationships or career/educational opportunities”), the GSMQ-9 only includes family conflicts for adolescents and study/work problems for adults. The reason for this is that our previous experience of research in this area and the pilot data collected while developing the GSMQ-9 show that these aspects are most indicative of problematic use of gaming or social media for the respective age groups. However, this could be seen as a limitation because it means that the GSMQ-9 does not fully capture this criterion as it is stated in the DSM-5 (APA,
2013). In sum, the present study provides initial support for the two-factor structure of the GSMQ-9 across two countries and among adolescents and young adults. However, additional studies are needed to validate the GSMQ-9 in other samples, including clinical samples. It would also be valuable to create a parent rating for use with younger children.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.