Heliyon 9 (2023) e14412

Contents lists available at ScienceDirect

Heliyon
journal homepage: www.cell.com/heliyon

Research article

A latent profile transition analysis and influencing factors of

internet addiction for adolescents: A short-term longitudinal study
Guangming Li
School of Psychology, Center for Studies of Psychological Application, And Guangdong Key Laboratory of Mental Health and Cognitive Science, South
China Normal University, Guangzhou 510631, China

A R T I C L E I N F O A B S T R A C T

Keywords: Internet addiction for adolescent, which is widely concerned by the whole society, has become a
Adolescent public health problem. Internet addiction not only had a negative impact on physical and mental
Internet addiction development of adolescents, but also was harmful to their study, life, interpersonal communi­
Latent profile analysis (LPA)
cation and personality formation, and so on. In recent years, the data analysis methods of lon­
Latent transition analysis (LTA)
gitudinal research have developed rapidly. It not only focused on the overall average growth
Latent profile transition analysis (LPTA)
Gender trend, but also considered the differences in the individual trends. Latent profile transition
Anxiety analysis (LPTA) is an extension of latent profile analysis (LPA) and latent transition analysis
Depression (LTA), and is a longitudinal data analysis method. LPTA can simultaneously estimate group
membership in multiple time points and their latent transition tendency among these subgroups
between each two time points. This study used LPTA to explore the development trend of
adolescent internet addiction over time and its influencing factors. 1033 adolescents participated
in a short-term 6-month longitudinal study with a total of three tests. Participants completed
internet addiction test, self-rating anxiety scale and self-rating depression scale. The results
showed that: (1) There are three categories of adolescent internet addiction, namely non-internet
addiction group, low-internet addiction group and high-internet addiction group. (2) Non-
internet addiction group has a strong stability. Low-internet addiction group has a high proba­
bility to become non-internet addiction group or high-internet addiction group. (3) Boys are more
likely than girls to develop into high-internet addiction group. Anxiety and depression both affect
the development of adolescent internet addiction.

1. Introduction

Internet Addiction Disorder (IAD) was originally proposed by Goldberg [1]. After that, Young et al. did a lot of research and
compiled corresponding measurement tools [2]. IAD refers to the enhancement of tolerance to network, withdrawal response,
persistent desire of networking and behavioral disorder due to prolonged inappropriate networking, doing harm to the individual’s
physical, psychological and social functions [3]. According to the report of adolescents networking surveyed by China Internet
Network Information Center up to the end of 2016, there were 195 million adolescents (under 25-year-old) using internet, accounting
for 50.7% of all netizen in China, and the prevalence rate of internet use for adolescents was 54.5%. Lots of adolescents are addicted to
the network while using it. This phenomenon has become a public health issue that the whole world is supported to pay attention to.
Meta-analysis showed that the incidence of IAD among adolescents from 2009 to 2014 ranged from 8%–12% [4]. Not only will internet

E-mail address: Lgm2004100@m.scnu.edu.cn.

https://doi.org/10.1016/j.heliyon.2023.e14412
Received 21 July 2022; Received in revised form 24 February 2023; Accepted 3 March 2023
Available online 9 March 2023
2405-8440/© 2023 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
G. Li Heliyon 9 (2023) e14412

addiction affect the physical and mental development of adolescents, but also do harm to their learning, life, interpersonal re­
lationships and personality formation [3,5–8].
The studies on internet addiction among adolescents in China mainly focus on cross-sectional research, mainly including its’
prevalence and influencing factors [9–11], but few studies have studied the development of internet addiction among Chinese ado­
lescents from the perspective of longitudinal analysis. Some longitudinal studies showed that adolescents’ internet addiction was
affected by family factors [12–14]. However, the researchers took adolescents as a whole to observe the development trend of internet
addiction over time and paid less attention to the individual developmental heterogeneity.
In recent years, the data analysis methods of longitudinal research have developed rapidly. It not only focused on the overall
average growth trend, but also considered the differences in individual trends [15]. Latent transition analysis (LTA) is an
individual-centered longitudinal data analysis method. It is a Lengthwise extension of latent Markov model based on Latent class
analysis (LCA). When an explicit and latent variable is classified as type data, LTA is a suitable analytical method. LTA can divide
subjects into different latent categories depending on different responses, in order to find out the development of latent categories over
time [16]. When the explicit variable is continuous data, the latent profile analysis (LPA) is used to explain the correlation between the
continuous explicit indexes by the latent categorical variables [17]. Accordingly, LTA is also extended to latent profile transition
analysis (LPTA), which can simultaneously estimate the latent categories at multiple time points and their development trends among
time points [18]. Previous studies have confirmed that LPTA can be used in longitudinal studies [18–20]. Therefore, it is of great
significance to explore the development rule of internet addiction with time through LPTA.
LCA and LPA are mainly aimed at latent classification of subjects, in which LCA is suitable for classified data and LPA is suitable for
continuous data. The main purpose of LTA and LTPA is to investigate the latent transition trend of latent classified subjects at different
time points; the former is for subjects classified by LCA, and the latter is for subjects classified by LPA.
Internet addiction for adolescents is affected by many factors, which can be generally divided into physical factors, social factors
and psychological factors. Internet addiction is similar to gambling addiction. With long-term surfing on internet, people would be
induced to secrete more dopamine in the brain, which will create a feeling of excitement and pleasure [21]. Social factors of internet
addiction include participants’ family environment, social culture and social life events. Some researches showed that internet
addiction is closely related to family factors of adolescents [10,21,22]. A bad family environment is extremely detrimental to the
growth of young people. Among psychological factors, adolescent personality traits, negative emotions, self-esteem, impulsiveness,
etc. all affect internet addiction [9,11,21,23]. Practical studies have shown that adolescent anxiety and depression can predict internet
addiction [23–25]. A two-way predictive effect was found between depression and internet addiction [26]. But whether anxiety and
depression can predict internet addiction over time is not yet clear. In addition, in the gender study of addictive behavior, boys are
more likely to be addicted to internet than girls [24,27,28]. Internet addiction has a greater relationship with emotional problems
(depression, anxiety) and indirect aggression among girls, and closely related to violations among boys [24].
At present, some researches on adolescents’ internet addiction have achieved some results, but there are also some shortcomings:
(1) Most researches use the total score of the scale to classify the subjects, ignoring the performance of the subjects on various items [2,
29]. This classification method does not mine the information in depth, so we should further explore whether there are individuals or
sub groups with different IAD performance patterns. (2) Most studies focus on horizontal research, mainly discussing the current
situation and influencing factors of teenagers’ internet addiction, but few studies explore the development of teenagers’ internet
addiction from the perspective of longitudinal analysis [9,22]. Little attention has been paid to the heterogeneity of individual
development, and there is no in-depth discussion on the development law of Internet addiction itself [13,30,31].
Latent profile transition analysis (LPTA) is an extension of latent profile analysis (LPA) and latent transition analysis (LTA). This
study used latent profile transition analysis, a longitudinal data analysis method, to explore the development trend of adolescent
internet addiction over time and its influencing factors.
Based on the above questions, this paper used LPTA to explore the development trend of adolescents’ internet addiction, and there
were two questions: (1) identifying different latent categories of adolescents’ internet addiction and their transitions; (2) exploring
whether anxiety, depression, and gender factors will affect the transition of adolescents’ internet addiction categories.

2. Method

2.1. Participants

The research sample used in this study is a purposive sampling technique, which means that the sample is taken with a specific
purpose based on the criteria set by the researcher. In the study, a total of 1423 middle school students were selected to participate in
the questionnaire. All students completed the questionnaire in the classroom. The questionnaire consisted of three parts, namely the
Internet Addiction Test, the Self-Rating Anxiety Scale and the Self-Rating Depression Scale, which were tested every 3 months. Three
tests were performed. At the first test, a total of 1423 students participated in the questionnaire, and the effective questionnaire was
1375 (effectiveness was 96.6%). After that, 1285 students participated in the second test, 1178 students participated in the third test,
and a total of 1033 participants participated in all three tests. Among them, there were 419 boys and 614 girls, 615 only children and
418 non only children; the average age of the subjects was 14.85 ± 1.84 years old.

2.2. Procedure

The middle school students from twenty classes were asked to evaluate three scales respectively. All these classes were selected

2
G. Li Heliyon 9 (2023) e14412

randomly. Experimenters had been trained before the survey was administrated to students.
The middle school students from twenty classes in three schools were randomly selected. Data were collected during the class using
a paper/pencil version survey administered to all students in these school classes. Students’ physiological indicators (e.g., height,
weight, blood pressure) were assessed within the first week following the baseline measures. The academic affairs offices of the three
schools provided the measurement sites. Students were wearing light clothes and were barefoot when measured at air-conditioned
rooms. Research staff was trained before they administered the survey. We gave parents the option to opt their child out of
completing and none of parents opted out and all the consents were informed written. Prior to study commencement, students’
consents by themselves were obtained, and this study was approved by the South China Normal University (SCNU) research ethics
board (Institutional Review Board).

2.3. Questionnaires

2.3.1. Internet addiction test

Internet Addiction Test (IAT) was compiled by Young to assess the Internet Addiction of middle school students [2]. The test
consists of 20 questions, using 5 points to score, 1 for never, 2 for a few, 3 for some time, 4 for usually, 5 for always. Test score ranges
from 20 to 100, the higher the score, the more addicted in surfing. When someone’s testing score no less than 50, he would be regarded
as internet addiction [32]. The Cronbach α coefficients of the test in three tests were 0.911, 0.915 and 0.925, respectively.

2.3.2. Self-rating anxiety scale

Anxiety was assessed using Zung’s Self-rating Anxiety Scale (SAS) [33]. SAS had 20 items (e.g., I feel more nervous and anxious
than usual), all of them rated on 4-point scales, ranging from “a little of the time” to “most of the time”. Items were reversely coded
whenever necessary. The Cronbach’s alpha coefficients of the anxiety scale were 0.762 at T1, 0.779 at T2, and 0.744 at T3.

2.3.3. Self-rating depression scale

Depression was measured using Zung’s Self-rating Depression Scale (SDS) [34]. SDS had 20 items (e.g., “I feel down-hearted and
blue.“), all of them rated on 4-point scales (1 = no or very little time; 2 = a small amount of time; 3 = a considerable amount of time; 4
= most or all of the time). Depression degree index was derived by dividing the sum of the scores on the 20 items by the maximum
possible score of 80 and expressed as a decimal. A score lower than 0.5 was recoded as ‘no depression’, 0.5–0.59 recoded as ‘moderate
to severe depression’, and 0.60–0.69 as ‘moderate to severe depression’, and higher than 0.69 recoded as ‘severe depression’ [35]. The
Cronbach’s alpha coefficients of the depression scale were 0.795 at T1, 0.824 at T2, and 0.845 at T3.

2.4. Data analysis

Descriptive analysis was performed using SPSS 21. Latent profile analysis was performed using Latent GOLD 5.1. After being
determined the optimal number of latent categories, latent transition analysis was performed with Mplus 8.0.

3. Results

3.1. Common method biases

There might be common method biases existed, because of the data collected by questionnaires. According to Harman single factor
test [36], the non-rotating principal component analysis was performed on the data of the three tests, respectively. The results show
that there are 13 common factors with eigen root greater than 1 at time point 1. The first factor explains the variation of 17.56%. At
time point 2, there are 12 common factors with eigen roots greater than 1, and the first factor explains the variation of 22.4%. At time
point 3, there are 11 common factors with eigenvalues greater than 1, and the first factor explains the variation of 17.79%. If the
non-rotating principal component factor analysis has a particularly large explanatory power, there may be a common method bias
[37], and there was no obvious common method bias in this study.

3.2. Adolescent internet addiction incidence

According to Table 1, the average score of internet addiction among adolescents at three time points gradually decreased with time,
and the incidence of internet addiction among adolescents gradually decreased with time.

Table 1
Descriptive statistics of adolescent internet addiction.
Statistics indicators Time point 1 Time point 2 Time point 3

Mean score 33.59 31.67 29.52

Standard deviation 12.01 10.83 10.23
Incidence of internet addiction 11.13% 8.03% 5.13%

3
G. Li Heliyon 9 (2023) e14412

3.3. Results of latent profile analysis

To explore the number of latent categories of internet addiction for Chinese adolescents, taking the 20 items of the scale as explicit
indicators, three time points’ internet addiction was analyzed separately by using latent profile analysis from 2 categories to 6 cat­
egories. The simulation results at three time points can be seen in Table 2.
In Table 2, LL is Log-likelihood Statistics. According to Table 2, at time point 1, the BIC, AIC, and SABIC indicators of the four-
category model were the smallest, indicating that it is optimal. In time point 2 and time point 3, as the number of latent categories
increases, the BIC, AIC and SABIC indicators became smaller and the parameters increase. After the number of latent categories in time
points 2 and 3 is 3, the magnitude of the decrease in the fit of the three models was gradually slowing down, so the number of latent
categories for selecting 3 categories as time points 2 and 3 is optimal. In order to keep the number of latent categories at three time
points consistent, considering the simplicity and accuracy of the model, the three-category model was finally selected as the best model
for three time points.
After confirming the three-category latent profile model as the final models of the three time points, three latent categories were
described and named for in-depth analysis. Figs. 1–3 present the mean distribution of scores for adolescent internet addiction at three
time points.
As can be seen from Fig. 1, the score trends of the three categories are consistent. According to the score, category 3 is named “high-
internet addiction group”, category 2 is named “low-internet addiction group”, the scores for most topics from category 1 are close to 2,
and 2 is " nearly never ", so category 1 is named “non-internet addiction group.” In Figs. 2 and 3, the three latent categories also show a
consistent trend, indicating the appropriateness of the three latent categories, so the naming of points 2 and 3 is consistent with time
point 1.

3.4. Results of latent transition analysis

Table 3 shows the latent state probabilities for the three latent categories at three time points, which is the proportion of each latent
category. At three time points, the non-internet addiction group had the highest latent state probability, followed by the low-internet
addiction group and the high-internet addiction group. With the passage of time, both the high-internet addiction group and the low-
internet addiction group showed a gradually decreasing trend, but the range of change was not large; the probability of the latent state
of the non-internet addiction group increased gently.
The latent profile transition analysis can longitudinally explore the changes among different latent states in three consecutive time
points, and analyze the development direction and stability of adolescent internet addiction from the perspective of transition
probability. For data tracking at three time points, there are two transition matrices, time point 1–2 and time point 2–3, respectively.
See Table 4 for the results.
As can be seen from Table 4, at time 1–2, the non-internet addiction group has strong stability, rarely changes, and the probability
of staying in its own category is 0.894. The probability of the low-internet addiction group staying in its own category is 0.536; the
probability of transitioning to a non-internet addiction group is 0.392. The probability of a high-internet addiction group staying in its
own category was 0.366, with a probability of 0.389 transitioning to a low-internet addiction group, with a probability of 0.245
transitioning to non-internet addiction group.
As can be seen from Table 4, at time 2–3, the non-internet addiction group still has high stability, the probability of staying in its
own category is 0.914, and rarely changed to the other two categories. The probability of the low-internet addiction group staying in its
own category was 0.503, comparing with time point 1–2, the low-internet addiction group has a higher probability of transitioning to
the non-internet addiction group. The probability of a high-internet addiction group staying in its own category was 0.320; high-

Table 2
Fitting indicators for the latent profile model.
Number of latent categories BIC(LL) AIC(LL) SABIC(LL) Parameter

Time point 1
2 38740.29 38340.13 38483.02 81
3 31310.83 30708.12 30923.34 122
4 25055.87 24250.61 24538.16 163
5 25559.37 24551.57 24911.44 204
6 21189.15 19978.80 20411.00 245
Time point 2
2 30616.28 30216.12 30359.02 81
3 19890.81 19288.10 19503.32 122
4 17258.81 16453.55 16741.10 163
5 14003.71 12995.91 13355.79 204
6 12077.73 10867.37 11299.58 245
Time point 3
2 19409.51 19009.36 19152.25 81
3 6708.83 6106.12 6321.34 122
4 4225.05 3419.79 3707.34 163
5 − 365.04 − 1372.84 − 1012.97 204
6 − 3129.28 − 4339.64 − 3907.43 245

4
G. Li Heliyon 9 (2023) e14412

Fig. 1. Mean score of 20 items of adolescent internet addiction at time point 1.

Fig. 2. Mean score of 20 items of adolescent internet addiction at time point 2.

Fig. 3. Mean score of 20 items of adolescent internet addiction at time point 3.

Table 3
Latent category probability of 3 time points.
Latent category Time point 1 Time point 2 Time point 3

Non-internet addiction group 0.612 0.688 0.753

Low-internet addiction group 0.310 0.253 0.211
High-internet addiction group 0.078 0.059 0.036

5
G. Li Heliyon 9 (2023) e14412

Table 4
Latent probability transition matrices of 3 time points.
Latent category High-internet addiction group Low-internet addiction group Non-internet addiction group

Time point 1-2

High-internet addiction group 0.366 0.389 0.245
Low-internet addiction group 0.072 0.536 0.392
Non-internet addiction group 0.013 0.093 0.894
Time point 2-3
High-internet addiction group 0.320 0.505 0.175
Low-internet addiction group 0.047 0.503 0.450
Non-internet addiction group 0.008 0.079 0.914

internet addiction group was more likely to transition to low-internet addiction groups comparing with time point 1–2.

3.5. Influencing factors of latent transition probability

Gender, anxiety and depression were included as covariates in the latent transition model. The effects of covariates on latent
transition probabilities were investigated by multiple logistic regression. The results of analysis were expressed by Odds Ratio (OR)
(see Table 5).
According to Table 5, the subjects in the high-internet addition group are taken as the reference group. Compared with boys, girls
have a higher probability of belonging to the low duration and low impact group, which is 2.24 times higher than the high duration and
high desire group. The OR of anxious subjects is less than 1, which shows that anxious teenagers are less likely to belong to non-internet
addition group or low-internet addition group, indicating that anxiety may increase teenagers’ internet addiction. However,
depression has no significant effect on IAD.
On the basis of exploring the effects of gender, anxiety and depression on the types of internet addiction in adolescents, multivariate
logistic regression analysis was used to further explore the impact of covariates on the transformation of the types of internet addiction
in adolescents. The odds ratio is a ratio of the probability changes of the category to which the subject belongs to in the transition
process, compared with a certain reference category. The odds ratio greater than 1 indicates that under the influence of the covariates,
the probability of the participants transitioning to other category increases, and less than 1 indicates that the probability of the
transition is reduced. The subjects in the non-internet addition group are taken as the reference group. The OR here refers to the ratio of
the probability of the subjects’ transition to the high internet addition group or the low internet addition group to the change of the
probability of the subjects’ transition to the non-internet addition group. See Tables 6–8 for the results.
Incorporating anxiety into the latent transition model, at time 1–2, with the high-internet addiction group at time point 2 as the
reference category, as the anxiety score increasing, the odds ratio of the three categories (non, low, high) at time point 1 changing into
(or staying at) non-internet addiction group (OR = 0.564, 0.579, 0.618) or low-internet addiction group was smaller (OR = 0.397,
0.429, 0.434). At time 2–3, with the high-internet addiction group at time point 3 as the reference category, as the anxiety score
increasing, the odds ratio of the three categories (non, low, high) at time point 2 changing into (or staying at) non-internet addiction
group (OR = 1.543, 1.535, 1.765) or low-internet addiction group was greater (OR = 4.791, 4.610, 4.689).
Incorporating depression into the latent transformation model, at time 1–2, with the high-internet addiction group at time point 2
as the reference category, as the depression score increasing, the odds ratio of the three categories (non, low, high) at time point 1
changing into (or staying at) non-internet addiction group (OR = 2.512, 2.440, 2.560) or low-internet addiction group was greater (OR
= 3.737, 3.702, 3.851). At time 2–3, with the high-internet addiction group at time point 3 as the reference category, as the anxiety
score increasing, the odds ratio of the three categories (non, low, high) at time point 2 changing into (or staying at) non-internet
addiction group (OR = 0.146, 0.149, 0.158) or low-internet addiction group was smaller (OR = 0.468, 0.456, 0.439).

4. Discussions

4.1. Classification of internet addiction for adolescents

The incidence of internet addiction among adolescents showed a gradual decline with time. The incidence of three time points
ranged from 5.13% to 11.13%, which was consistent with previous studies [4,8,11]. There were three latent categories of adolescent
internet addiction as the result of LPA, namely, non-internet addiction group, low-internet addiction group and high-internet addiction

Table 5
The OR of T1 latent state probability with covariates.
Influencing factors High-internet addiction group Low-internet addiction group Non-internet addiction group

Gender REF 1.31 2.24*

Anxiety REF 0.95* 0.90***
Depression REF 1.01 1.01

Note：*p ＜ 0.05，***p ＜ 0.001. REF is reference group.

6
G. Li Heliyon 9 (2023) e14412

Table 6
The OR of transition probability with covariates gender.
Influencing factors Latent status High-internet addiction group Low-internet addiction group Non-internet addiction group

T1-T2
Gender High-internet addiction group REF 0.927 0.127
Low-internet addiction group REF 1.030 0.098
Non-internet addiction group REF 2.142 0.099
T2-T3
Gender High-internet addiction group REF 2.620 0.574
Low-internet addiction group REF 2.012 0.232
Non-internet addiction group REF 2.067 0.969

Incorporating gender into the latent transition model, at time 1–2, with the high-internet addiction group at time point 2 as the reference category,
boys’ odds ratio, comparing girls’, of the three categories (non, low, high) at time point 1 changing into (or staying at) non-internet addiction group
was smaller (OR = 0.127, 0.098, 0.099); the probability of low-internet addiction group transitioning to high-internet addiction group was small (OR
= 0.927), and non-internet addiction groups and low-internet addiction groups were more likely to undergo transitions (OR = 2.142 and 1.030). At
time 2–3, with the high-internet addiction group at time point 3 as the reference category, boys’ odds ratio, comparing girls’, of the three categories
(non, low, high) at time point 2 changing into (or staying at) non-internet addiction group was smaller (OR = 0.574, 0.232, 0.969), the probability of
transitioning to a low-addiction group was greater (OR = 2.620, 2.012, 2.067).

Table 7
The OR of transition probability with covariates anxiety.
Influencing factors Latent status High-internet addiction group Low-internet addiction group Non-internet addiction group

T1-T2
Anxiety High-internet addiction group REF 0.397 0.564
Low-internet addiction group REF 0.429 0.579
Non-internet addiction group REF 0.434 0.618
T2-T3
Anxiety High-internet addiction group REF 4.971 1.543
Low-internet addiction group REF 4.610 1.535
Non-internet addiction group REF 4.689 1.765

Table 8
The OR of transition probability with covariates depression.
Influencing factors Latent status High-internet addiction group Low-internet addiction group Non-internet addiction group

T1-T2
Depression High-internet addiction group REF 3.737 2.512
Low-internet addiction group REF 3.702 2.440
Non-internet addiction group REF 3.851 2.560
T2-T3
Depression High-internet addiction group REF 0.468 0.146
Low-internet addiction group REF 0.456 0.149
Non-internet addiction group REF 0.439 0.158

group. At three time points, the proportion of participants in non-internet addiction group was the highest, the proportion of par­
ticipants in high-internet addiction group was the lowest, and the proportion of participants in low-internet addiction group was
between the two. With time going by, the proportion of participants in the non-internet addiction group gradually increased, while the
participants in the low-internet addiction group and the high-internet addiction group gradually decreased. Preceding studies divided
adolescents into two groups, namely normal group and internet addiction group. The difference between two groups was obvious. The
networking time of the participants in internet addiction group was about 3 times as the non-internet group [38]. This study validated
that there is different categories in internet addicted adolescents, and found low-internet addiction group, a new category of adolescent
internet addiction. Each type of internet addiction has its characteristics. The high-internet addiction group’s score of each item was
higher, especially the score of item 1, 2, 5, 10, 11 and 16, reflecting the participants surf long time on internet, long for networking, and
would ignore what they have to do because of networking. This was consistent with the two dimensions of the “China Youth Network
Problem Spectrum” [22]. In terms of network usage, high-internet addiction groups would devote more time in the network; in terms
of the severity of the internet addiction, the network using has been affected high-internet addiction group’s daily lives. In addition,
networking time is a significant predictor of internet addiction [38]. In item 8, 19, 20, non-internet addiction group scored the lowest
point, reflecting normal networking would not take negative effect to adolescents’ daily life, interpersonal relation and emotional
changes. Score of low-internet addiction group was between the score of high-internet addiction group and non-internet addiction
group. Though participants of low-internet addiction group have the tendencies of internet addiction and longer time of networking,
their daily life, study and interpersonal relationship were affected negatively by networking.

7
G. Li Heliyon 9 (2023) e14412

4.2. Transition of internet addiction for adolescents

With time going by, internet addiction of adolescents showed a trend of getting better. As far as the LPTA results are concerned, the
non-internet addiction group had strong stability and rarely changes. Low-internet addiction group had about half the probability of
staying in its own category and was more likely to change into non-internet addiction group. High-internet addiction group had a low
probability of staying at its category, and was easier to become low-internet addiction group. Therefore, in the development process of
adolescent internet addiction, it is important to intervene in high-internet addicts promptly and pay attention to low-internet addicts.
At present, intervention and treatment of internet addiction mainly include psychological behavior therapy and drug therapy. Among
them, cognitive-behavior therapy, group counseling, family therapy and other psychological therapy can intervene and prevent
internet addiction [3,39].
After incorporating gender into latent profile transition model, in two transition, boys, comparing with girls, had a higher prob­
ability to become members of high-internet addiction group. It was consistent with previous studies, that boys are easier to get internet
addiction than girls [24,26–28]. The gender difference will exist over time; it may be affected by the difference of surfing behaviors and
emotion between boys and girls.

4.3. Influencing factors of internet addiction for adolescents

To investigate how anxiety and depression effect the development of adolescent internet addiction, with the high-internet
addiction group as the reference group, the results showed that, as the anxiety increasing, the high-internet addiction group had a
smaller probability to become low-internet addiction group and non-internet addition group at time point 1–2, while had a greater
probability to become low-internet addiction group and non-internet addition group at time point 2–3. Past studies showed that
anxiety positively predicts internet addiction [23,25,40]. On the one hand, adolescents will be addicted to the internet in order to
alleviate anxiety. On the other hand, the adverse effects of internet addiction on young people’s learning, life and interpersonal re­
lationships will make them more anxious. Similarly, with the high-internet addiction group as the reference category, as the depression
increasing, there was a greater possibility at time point 1–2 to change to non-internet addiction group and low-internet addiction
group, and at time point 2–3 there was a smaller possibility of transitioning to the non-internet addiction group and the low-internet
addiction group. It may be because the occurring of depression is a slow process, and the melancholy spirit and loss of interest are
gradually occurring. Adolescents are less sensitive in the initial stages of depression and are less likely to become high-internet addicts.
As depression continues to increase, adolescents will spend more time surfing the internet in order to alleviate depression, and more
likely to become high-network addicts. Some scholars have found a two-way relationship between depression and internet addiction
through cross-lagged regression analysis [26]. Therefore, negative emotions, such as anxiety and depression, not only can predict
adolescents’ internet addiction [11,40], but also would affect development of the negative emotions.
Therefore, gender, anxiety and depression are all predisposing factors for adolescents’ internet addiction. Adolescents should
strengthen the recognition and regulation of negative emotion, which cannot be alleviated by indulging in the network. Educators can
use a variety of psychological behavioral therapies in combination, promptly intervene in high- and low-internet addicts, and use
clinical medication if necessary. Of course, the impact of anxiety and depression on the development of adolescent internet addiction is
only the exploration, and it needs to be verified by subsequent research.
In the study, 1033 adolescents participated in a short-term 6-month longitudinal study with a total of three tests. Participants
completed internet addiction test, self-rating anxiety scale and self-rating depression scale. LPTA and two latent variable models (latent
profile models and latent transition models) are used. First of all, a latent profile analysis was conducted on adolescent internet
addiction items. According to the model fitting indicators, a latent profile model with the optimal latent class number was selected. The
model fitting indicators mainly use the information evaluation indicator AIC, BIC and SABIC. When the model is compared, the smaller
the values of three indicators, the better the model fit, and the more parameters, the more complex the model. Secondly, based on the
optimal number of latent classes, the three time points’ items were simultaneously included in a latent transition model to analysis.
Thirdly, gender, anxiety, and depression were taken as covariates and included in latent transition model respectively. Multivariate
logistic regression was used to explore the influence of covariates on transition probability.
The innovation of this study is to use the method of LPTA to explore the development trend and influencing factors of adolescents’
internet addiction from the perspective of longitudinal research. Firstly, compared with similar longitudinal studies, the focus has
shifted from the overall trend to the heterogeneous development of sub groups, and more attention has been paid to the transformation
of different latent categories of adolescent Internet addiction. Secondly, a new longitudinal data analysis method, namely LPTA, is
adopted, and the observed explicit indicators are expanded from category variables to continuous variables. This method can be widely
used in the longitudinal research of continuous data. Finally, the research results show that there are three latent categories of ado­
lescents’ IAD, namely, non-internet addiction group, low-internet addiction group and high-internet addiction group. Without any
intervention, the non-internet addiction group has strong stability. Nonetheless, the low-internet addiction group is more likely to be
transformed into the non-internet addiction group, and the high-internet addiction group is more likely to be transformed into the low-
internet addiction group. Therefore, it is necessary to timely intervene the low-internet addiction group and high-internet group to
promote their development in a better direction.

4.4. Limitations

Although this study discussed the development trend and influencing factors of internet addiction for adolescents from the

8
G. Li Heliyon 9 (2023) e14412

perspective of longitudinal study, there are still many limitations. First of all, due to the short time interval of the survey, only a 6-
month follow-up was carried out to obtain the short-term development trend of adolescent internet addiction, which affected the
popularization of the research results. A short-term time span (i.e., nine months in this study, with a 3-month interval between data
time points) may have limited the capacity to precisely capture the developmental trajectory courses for adolescents. Secondly, the
subjects of this study are middle school students, but college students are not involved. Nonetheless, in some researches about internet
addiction, many researchers classify college students as adolescents. In addition, the direction of change of different types of internet
addiction for adolescents and the role of influencing factors were discussed for the first time, which needs to be further verified by
subsequent studies. Finally, although this study has some limitations, it is of guiding significance for clinical prevention and treatment
to identify different types of adolescent internet addiction and their development patterns over time. The latent profile shift analysis
method adopted in this study can be used in longitudinal studies to provide a new analytical idea for the development of internet
addiction for adolescents.

5. Conclusions

LPTA showed that there were three types of internet addiction among adolescents, namely, non-internet addiction group, low-
internet addiction group, and high-internet addiction group. The non-internet addiction group had strong stability and was un­
likely to change, but the probability that the low-internet addiction group turned into non-internet addiction group and high-internet
addiction group turned into low-internet addiction group was higher. The effects of covariates on adolescent internet addiction showed
that boys were more likely than girls to develop into high-internet addict and anxiety and depression both affected the development of
adolescent internet addiction. It is concluded that with the development of internet addiction among adolescents, it is more important
to promptly intervene in high-internet addiction group and focus on low-internet addiction group. Gender, anxiety and depression
were all susceptibility factors, so adolescents should strengthen the identification and regulation of negative emotion, which cannot be
relieved simply by indulging in the internet. Educators can integrate a variety of psychological behavioral therapies to prevent and
intervene in adolescent internet addiction.

Author contribution statement

Guangming Li: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data;
Contributed reagents, materials, analysis tools or data; Wrote the paper.

Funding statement

This work was supported by the Characteristic Innovation Project of Colleges and Universities in Guangdong Province (Philosophy
and Social Science of Educational Science) [2021wtscx020] and the Natural Science Foundation of Guangdong Province
[2021A1515012516].

Data availability statement

Data included in article/supp. material/referenced in article.

Declaration of interest’s statement

The authors declare no competing interests.

References

[1] N.A. Gonzalez, Internet Addiction Disorder and its Relation to Impulse Control. Doctoral Dissertation, Texas A &M University-Kingsville, 2002.
[2] K.S. Young, C.N. De Abreu, Internet Addiction: A Handbook and Guide to Evaluation and Treatment, John Wiley & Sons, 2010.
[3] X. Fang, L. Liu, L. Deng, Q. Liu, W. Su, J. Lan, The prevention and intervention on adolescent internet addiction, Psychol. Dev. Educ. 31 (1) (2015) 100–107.
[4] H. Bian, Y. Liu, N. Li, T. Liu, Meta Analysis for internet addiction rate among adolescents, Chin. J. Sch. Heal. 37 (1) (2016) 67–70.
[5] J.Y. Geng, L. Han, F.Q. Gao, M. Jou, C.C. Huang, Internet addiction and procrastination among Chinese young adults: a moderated mediation model, Comput.
Hum. Behav. 84 (2018) 320–333.
[6] L.T. Lam, Internet gaming addiction, problematic use of the internet, and sleep problems: a systematic review, Curr. Psychiatr. Rep. 16 (4) (2014) 444.
[7] L. Milani, D. Osualdella, B.P. Di, Quality of interpersonal relationships and problematic Internet use in adolescence, Cyberpsychol. Behav. 12 (6) (2009)
681–684.
[8] M.M. Spada, An overview of problematic Internet use, Addict. Behav. 39 (1) (2014) 3–6.
[9] J. He, P. Zhu, Y. Nie, S. Ying, The influence of personality on internet addiction and its psychological mechanisms: a review, Chin. J. Clin. Psychol. 25 (2) (2017)
221–224.
[10] M. Jiang, Y. Wan, Z. Cheng, Influence of family rearing patterns on internet addiction and mediating effect of personality, Chin. J. Clin. Psychol. 25 (5) (2017)
907–910.
[11] F. Zhou, Y. Liu, M. Guo, S. Jiang, Negative affect of adolescents and internet addiction modulating effect of orientations to happiness, Chin. J. Clin. Psychol. 25
(2) (2017) 208–212.
[12] C.H. Ko, P.W. Wang, T.L. Liu, C.F. Yen, C.S. Chen, J.Y. Yen, The bidirectional associations between the family factors and internet addiction among adolescents
in a prospective investigation, Psychiatr. Clin. Neurosci. 69 (4) (2014) 192–220.

9
G. Li Heliyon 9 (2023) e14412

[13] D.T.L. Shek, X.Q. Zhu, C.M.S. Ma, The influence of parental control and parent-child relational qualities on adolescent internet addiction: a 3-year longitudinal
study in Hong Kong, Front. Psychol. 9 (2018) 1–16.
[14] R.J.J.M. van Den Eijnden, S. Renske, A.A. Vermulst, T.J. van Rooij, R.C.M.E. Engels, Compulsive internet use among adolescents: bidirectional parent-child
relationships, J. Abnorm. Child Psychol. 38 (1) (2010) 77–89.
[15] H. Liu, Meng Qi, A review on longitudinal data analysis method and it’s development, Adv. Psychol. Sci. 11 (5) (2003) 586–592.
[16] B. Wang, M. Zhang, J. Zhang, J. Hu, Describing the stage process of individual through transition matrix: latent transition model, Psychol. Res. 8 (4) (2015)
36–43.
[17] M. Wang, Q. Deng, X. Bi, H. Ye, W. Yang, Performance of the entropy as an index of classification accuracy in latent profile analysis: a Monte Carlo simulation
study, Acta Psychol. Sin. 49 (11) (2017) 1473–1482.
[18] L.K. Fryer, (Latent) transitions to learning at university: a latent profile transition analysis of first-year Japanese students, High Educ. 73 (3) (2017) 519–537.
[19] E. Bourchtein, J.M. Langberg, J.S. Owens, S.W. Evans, R.A. Perera, Is the positive illusory bias common in young adolescents with ADHD? A fresh look at
prevalence and stability using latent profile and transition analyses, J. Abnorm. Child Psychol. 45 (6) (2017) 1063–1075.
[20] M. Flaig, B.A. Simonsmeier, A.K. Mayer, T. Rosman, J. Gorges, M. Schneider, Conceptual change and knowledge integration as learning processes in higher
education: a latent transition analysis, Learn. Indiv Differ 62 (2018) 49–61.
[21] S. Liu, Z. Zhang, The social-paychological-physiological model of internet addiction and its research trends, Chin. J. Appl. Psych. 10 (2) (2004) 48–54.
[22] W. Gao, Z. Chen, A study on psychopathology and psychotherapy of internet addiction, Adv. Psychol. Sci. 14 (4) (2006) 596–603.
[23] C.F. Yen, W.J. Chou, T.L. Liu, P. Yang, H.F. Hu, The association of Internet addiction symptoms with anxiety, depression and self-esteem among adolescents with
attention-deficit/hyperactivity disorder, Compr. Psychiatr. 55 (7) (2014) 1601–1608.
[24] Q. Zhang, Y. Wang, C. Yuan, X. Zhang, Y. Li, The gender effect on the relationship between internet addiction and emotional and behavioral problems in
adolescents, Chin. J. Clin. Psychol. 22 (6) (2014) 1004–1009.
[25] E. Dalbudak, C. Evren, The relationship of internet addiction severity with attention deficit hyperactivity disorder symptoms in Turkish university students;
impact of personality traits, depression and anxiety, Compr. Psychiatr. 55 (3) (2014) 497–503.
[26] S. Xun, Z. Huang, F. Guo, J. Hou, Z. Chen, Bidirectional relationship between internet addiction and depression in adolescents, Chin. J. Clin. Psychol. 21 (4)
(2013) 613–615.
[27] M. Baloglu, H.I.O. Kozan, S. Kesici, Gender differences in and the relationships between social anxiety and problematic internet use: canonical analysis, J. Med.
Internet Res. 20 (1) (2018) 1–9.
[28] C.H. Ko, J.Y. Yen, C.C. Chen, S.H. Chen, C.F. Yen, Gender differences and related factors affecting online gaming addiction among Taiwanese adolescents,
J. Nerv. Ment. Dis. 193 (4) (2005) 273–277.
[29] R. Kojima, R. Shinohara, Y. Akiyama, H. Yokomichi, Z. Yamagata, Temporal directional relationship between problematic internet use and depressive symptoms
among Japanese adolescents: a random intercept, cross-lagged panel model, Addict. Behav. 120 (2021) 1–6.
[30] H. Chen, J. Wang, Y. Lin, S. Yang, Association of internet addiction with family functionality, depression, self-efficacy and self-esteem among early adolescents,
Int. J. Environ. Res. Publ. Health 17 (23) (2020) 1–13.
[31] D. Luo, L. Tan, D. Shen, Z. Gao, L. Yu, M. Lai, J. Xu, J. Li, Characteristics of depression, anxiety, impulsivity, and aggression among various types of drug users
and factors for developing severe depression: a cross-sectional study, BMC Psychiatr. 22 (2022) 1–12.
[32] X. Wu, Study on Interpersonal Characteristics of Internet Addiction Among Adolescents and its Comprehensive Intervention. Unpublished Doctoral Dissertation,
Huazhong University of Science and Technology, 2013.
[33] W.W. Zung, A rating instrument for anxiety disorders, Psychosomatics 12 (6) (1971) 371–379.
[34] W.W. Zung, A self-rating depression scale, Arch. Gen. Psychiatr. 12 (1) (1965) 63–70.
[35] C. Zhang, Y. Xue, H. Zhao, Prevalence and related influencing factors of depressive symptoms among empty-nest elderly in Shanxi, China, J. Affect. Disord. 245
(14) (2019) 750–756.
[36] P.M. Podsakoff, S.B. Mackenzie, J.Y. Le, N.P. Podsakoff, Common method biases in behavioral research: a critical review of the literature and recommended
remedies, J. Appl. Psychol. 88 (5) (2003) 879–903.
[37] H. Zhou, L. Long, Statistical remedies for common method biases, Adv. Psychol. Sci. 12 (6) (2004) 942–950.
[38] S. Zhou, Z. Tang, Y. Peng, Internet-related behavior characteristics of adolescents with internet addiction, Chin. J. Clin. Psychol. 17 (2) (2009) 151–153.
[39] F. Wang, M. Ni, H. Li, J. Wo, The diagnosis and treatment of internet addiction, Psychol. Sci. 31 (3) (2008) 714–716.
[40] Y. Xu, Y. Yang, D. Wu, Mediating effect of trait impulsivity between negative affect and internet overuse in college students, Chin. J. Clin. Psychol. 21 (6) (2013)
942–945.

10