Participants
Data were collected between July 2017 and January 2020 in the LIFE Child Study Center in Leipzig, Germany. The LIFE Child study is a cohort study aiming to monitor child development from pregnancy to early adulthood [38]. Participants are mainly recruited via advertisements at different institutions, e.g., schools or Health Centers. All children and adolescents interested in the study and not suffering from chronic or syndromal diseases are allowed to participate.
For the present project, only cross-sectional analyses were performed. In the case that children had participated at more than one time point during the period of data acquisition (n = 332), only data collected at the last study visit were considered. The final sample comprised 664 adolescents (350 boys and 314 girls) aged between 10 and 18 years (M = 14.00, SD = 2.13).
The LIFE Child study was designed in accordance with the Declaration of Helsinki, and the study protocol was approved by the Ethics Committee of the Medical Faculty of the University of Leipzig (Reg. No. 264/10-ek). All parents provided informed written consent before participation of their children. From the age of 12 years on, children also gave informed written consent.
Measurements
Data on mobile phone and computer use, behavioral strengths/difficulties and school grades were assessed using questionnaires completed by the adolescents themselves. Their parents provided information on socioeconomic status (SES).
Media use: The media questionnaire used here was designed in 2017 by researchers of the LIFE Child study and was already used in another study project [39]. The English version of the questionnaire is shown in Additional file 1. The questionnaire assesses the use of different media devices, namely TV, games console, computers (including personal computers, tablets, laptops), and mobile phones (including smartphones and other mobile phones). As the activities a device might be used for were assessed for computer and mobile phone use only, the present analyses focus on the use of these media devices. A first block of questions assessed the time spent with each device. Separate questions were asked for the use on weekends and weekdays and for online and offline use of either mobile phone or computer. The following response categories could be chosen: never, 30 minutes, 1-2 hours, 3-4 hours, more than 4 hours. A second block of questions assessed the specific activities mobile phones or computers might be used for. Children indicated how often (in relation to the total time spent with the device (computer or mobile phone)) they usually spend time with each of the following activities: a) gaming (excluding learning games), b) playing learning games, c) searching for information, d) social media use, e) traditional communication, and f) watching videos. Tradition communication included telephone calling or writing text messages (via mobile phone se) or mailing (via computer). Response categories were never, sometimes, and often.
Behavioral strengths and difficulties: Behavioral strengths and difficulties of the participants were assessed using the German version of the Strength and Difficulties Questionnaire (SDQ) [40]. This screening instrument consists of five scales, each represented by five items. Four of these scales assess behavioral difficulties (emotional problems, conduct problems, symptoms of hyperactivity/inattention, peer-relationship problems). One scale assesses the strength prosocial behavior. The score in each scale ranges between 0 and 10, with higher scores indicating more behavioral difficulties/strengths.
School grades: Participants reported on their school grades in mathematics and German (first language) as documented on the most recent school record. In Germany, school grades range from 1 to 6, with a lower number indicating better performance (1 = “very good”, 2 = “good”, 3 = “satisfactory”, 4 = “sufficient”, 5 = “deficient”, 6 = “insufficient”).
SES: The SES of all study participants was assessed by a composite score (Winkler index). This score combines information on parental education, parental occupation, and the household equivalent income [41]. It ranges between 3 and 21, with higher scores indicating higher SES. In a large representative German sample [41], cut-offs were created to distinguish low (lowest 20%), middle (middle 60%), and high SES families (highest 20%) based on the score. These cut-offs were used to categorize the SES of participating families accordingly.
Data analysis
All analyses were performed using R [42]. In order to better evaluate the duration of participants’ computer and mobile phone use, response categories of the corresponding questions were transformed into durations (never = 0, 30 minutes = 0.5, 1-2 hours = 1.5, 3-4 hours = 3.5, more than 4 hours = 5), and offline and online use were summed up. Furthermore, usage times on weekends and weekdays were averaged ((usage on weekdays*5 + usage on weekends*2)/7).
To estimate the time spent with the different media activities, information on the total duration of mobile phone or computer use and the frequency of media activities were combined. The scores for the frequencies of the single media activities were assigned weights (“never” = 0, “sometimes” = 1, “often” = 2) and the estimated time per activity was calculated as follows: Total mobile phone (or computer) usage time*weight for the specific activity/sum of all weights per child. Finally, the durations of the single media activities via mobile phone and computer were summed up (e.g., social media use = social media use via mobile phone + social media use via computer).
Interrelations between the different media activities were assessed using Spearman correlation coefficients. Associations of the different media activities with behavioral strengths and difficulties were investigated using linear regression analyses. The scores on the different scales of the SDQ or school grades were included as dependent variables, and the durations of the different media activities were included as independent variables. In a first step, separate analyses, i.e., one model per media activity, were performed. In a second step, all activities were included in one model. All associations were indicated by non-standardized regression coefficients. Age, gender, and SES (low versus middle versus high) were included as control variables in all models.