Adaptation and Psychometric Analysis of the Test of Mobile Phone Dependence—Brief Version in Italian Adolescents

Since the diffusion of recent models of mobile phones, anyone with an internet connection can communicate continuously and search for information. This raises some questions about the possible consequences of problematic mobile phone use (PMPU) in a complex life phase such as adolescence. Therefore, we performed a psychometric analysis of the brief version of the Test of Mobile Phone Dependence (TMD) in Italy. The sample comprised 575 Italian adolescents aged 11 to 18 years. Data were collected using the TMD-brief, the Personality Inventory for the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and the Patient-Reported Outcomes Measurement Information System (PROMIS) Sleep Disturbance Short Form. Regarding test dimensionality, the best-fit measurement model included four factors: “Abstinence”; “Abuse and interference with other activities”; “Tolerance”; and “Lack of control” (Satorra–Bentler χ2 (48) = 185.96, p < 0.01; robust root mean square error of approximation (RMSEA) = 0.079 (90% confidence interval (CI): 0.067; 0.091); robust TLI = 0.904; robust comparative fit index (CFI) = 0.930). The Italian version of the TMD-brief was found to have good reliability and psychometric properties, and a four-factorial structure. PMPU predicted significant sleep disturbances and this relationship was moderated by clinical personality traits. Findings from this study support the use of the Italian version of the TMD-brief as a screening tool to investigate PMPU in Italian adolescents.


Introduction
More than 4 billion people (60% of the world's population) are connected to the internet [1,2]. In the last two decades, a real technological and informational revolution has taken place. The internet influences every aspect of people's daily life, directly or indirectly, particularly with respect to communication and interaction between individuals. The International Telecommunication Union [3] estimates indicate that 70% of the world's young people (15 to 24 years old) are online, and this proportion is significantly higher compared to that of the total population using the internet.
The possibility of continuous communication and connection goes beyond the natural concept of time and physical distance. Nevertheless, we can be in real-time communication with our interlocutor by a click of the screen and see him through a video chat even if we are physically separated by great distances. Essentially, smartphones makes a constant connection to the internet and communication with others possible. In 2019, 90% of the world's population who accessed the internet did it through their mobile phones [2].
At the same time, this continuous connection influences concepts of private place, intimacy and privacy. Worldwide, 3.8 billion people use social media, representing 49% of the world's population [2]. Social networks allow us to have our own profile or internet link between PMPU and dysfunctional behaviors and psychopathological symptoms such as fear of missing out (FoMO) [29], hyperactivity, poor prosocial behavior [28], low selfesteem, stress, anxiety and depression [30] and suicidal ideation and suicide attempts [19]. Recently, the Italian Pediatric Society has focused attention on the negative influences of PMPU and other technological instruments on the psychophysical development of adolescents, providing advice for parents and clinicians [31].
Research has shown the potential effects of mobile phone use on sleep quality and disturbances. Sleep problems were related to both smartphone use [32] and PMPU [33,34]. The use of smartphones, especially during bedtime, is negatively associated with different sleep characteristics among children and adolescents, such as inadequate sleep quantity, poorer quality and high daytime sleepiness [32]. These findings raise concerns as sleep plays an important role in maintaining good psychophysical health. Despite the fact that some sleep characteristics go through a period of change in adolescence, previous studies have suggested that sleep problems can have an impact on several aspects of daily functioning such as difficulties in regulating negative emotions and externalizing problems [35][36][37].
At the same time, the relationship between PMPU and sleep quality and disturbances may vary depending upon specific characteristics [38]. In other words, certain adolescents may be more susceptible to the effects of PMPU. Pathological personality traits are related to impairment in both the interpersonal and personal context as well as with addictive behaviors and problematic use of technology, including PMPU [39]. Few studies have examined the role of personality traits as moderators of the associations between the problematic use of technology and daily functioning among adolescents [40] while they are lacking for PMPU. Therefore, we examined whether the direct effect of PMPU on sleep quality and disturbances was moderated by personality traits as conceptualized by the Alternative Model of Personality Disorder provided in Section 3 of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [41]. According to the Alternative Model, five broad domains of maladaptive personality traits reflect phenotypic variability in personality dysfunction. These five major domains include: Negative Affectivity (frequent and intense experiences of a wide range of negative emotions); Detachment (avoidance of socioemotional experience, including both withdrawal from interpersonal interactions and restricted affective experience and expression); Antagonism (behaviors that put one at odds with other people, including ideas of self-importance and callousness towards and/or unawareness of others); Disinhibition (orientation towards immediate gratification and impulsivity with no regard for past learning or future consequences); and Psychoticism (exhibiting culturally incongruent, odd, eccentric, or unusual behaviors and cognitions, both in content and process). The present study had an exploratory character considering that no previous study examined the moderator role of personality traits of the link between PMPU and sleep disturbances.
In light of the above, the present study aimed to adapt and validate a multicultural measure assessing PMPU in an Italian population, the Brief Multicultural Version of the Test of Mobile Phone Dependence (TMD-brief) questionnaire [12], as well as to analyze its psychometric properties. Choliz et al. [12] have identified a factorial structure that represents the main components of addiction: Abstinence, Abuse and interference with other activities, Tolerance and Lack of control. This factorial structure was replicated for different samples included in their study and the TMD-brief was sufficiently sensitive to detect differences among several populations with respect to PMPU. To the best of our knowledge, this is the first study to examine this topic using multiculturally validated instruments for PMPU in a population of Italian adolescents. In particular, the specific goals were: (a) to adapt a version of the TMD-brief using a translation-back-translation method in a sample of Italian participants aged 11 to 18 years old; (b) to analyze its dimensional structure using confirmatory factor analysis; and (c) to investigate the relationship between symptoms of PMPU and daily functioning, namely, personality functioning and sleep disturbances.

Participants and Procedure
A sample of 575 adolescents (294 males and 277 females) aged 11 to 18 years with an average age of approximately 14.1 (standard deviation (SD) = 1.86; males mean age = 14.3, SD = 1.89; female mean age = 13.9, SD = 1.81) took part in the study. Due to missing responses, 47 participants (approximately 8%) were excluded from the study. The final sample comprised 528 adolescents.
Inclusion criteria for participation in this study included the use or possession of a mobile phone or smartphone and the ability to understand the written Italian language. Exclusion criteria included a history of significant neurological illness or brain injury and the use of medications that could affect the study's outcomes.
Participants were recruited from two secondary schools in central Italy. This was a convenience sample and the schools do not represent a random sample of the school population. The aim of the study was illustrated to the headmasters and teachers of each school, outlining that the study was designed to evaluate the psychophysical health of students. Informed consent was obtained from both participants who voluntarily participated in the study and their parents before inclusion in the study. After their consensus, questionnaires were briefly presented in the classrooms during school time. The administration lasted approximately 30 min and participants filled out the questionnaires collectively in the classroom. Confidentiality of participants was ensured by assigning each student a code with no identifiable features. Every precaution was taken to protect the privacy of research subjects and the confidentiality of their personal information. This study was approved by the Ethics Committee of the Department of Dynamic and Clinical Psychology, Sapienza University of Rome (Approval Number: 8/2017).

Measures
All participants completed the following questionnaires. The TMD-brief by Choliz et al. [12] is a 12-item questionnaire assessing four latent constructs related to Mobile Phone Dependence: Abstinence ("If my mobile phone were broken for an extended period of time and took a long time to fix, I would feel very bad"; "If I don't have my mobile phone, I feel bad", "I don't think I could stand spending a week without a mobile phone"); Abuse and interference with other activities ("I spend more time than I would like to talking on my mobile phone, sending SMSs, or using WhatsApp", "I have gone to bed later or slept less because I was using my mobile phone", "I use my mobile phone (calls, SMSs, WhatsApp...) in situations where, even though not dangerous, it is not appropriate to do so (eating, while other people are talking to me, etc.)"); Tolerance ("I need to use my mobile phone more and more often", "When I have my mobile phone with me, I can't stop using it", "Since I got my mobile phone, I have increased the number of SMSs I send"); and Lack of Control ("As soon as I get up in the morning, the first thing I do is see who has called me on my mobile phone or if someone has sent me an SMS", "When I feel lonely, I use my mobile phone (calls, SMSs, WhatsApp...)", "I would grab my mobile phone and send a message or make a call right now"). Following the forward-back-translation procedure, all items were translated into Italian by a psychologist mother-tongue expert. The final version of the TMD-brief was then back-translated from Italian to English by an English mother-tongue expert with a degree in Psychology. A comparison of the original and back-translated version showed that no further modifications were needed. The Italian version of the TDM-brief (TDM-brief-Ita) is presented in Appendix A. Before running a confirmatory factor analysis (CFA), we conduct a preliminary exploratory factor analysis (principal components) by extracting four (oblimin) rotated factors. Results showed that the first 4 components explained approximately 72% of the total observed variance. On the first extracted component (29% explained variance), all Abstinence items saturated highly (>0.40). Similarly, on the second extracted component (20% explained variance), all Abuse items saturated highly. However, items expected to load on the third (Tolerance) and fourth (Lack of Control) components reported no clear pattern. To this aim, we decided to run a Confirmatory Factor Analysis in order to consider alternative factor structures.
The Personality Inventory for DSM-5 Brief Form (PID-5-BF) [41] investigates Negative Affect, Detachment, Antagonism, Disinhibition and Psychoticism. The PID-5-BF includes 25 items, five for each domain, with four response alternatives distributed on a four-point Likert scale, from 0 (very false/often false) to 3 (very true/often true). The Italian validation study showed good psychometric properties [42]. Average scores for each domain are considered in the present study. Reliability estimates in the present sample for the personality inventory domains were: Negative Affect Cronbach alpha = 0.  [43] is an 8-item self-report questionnaire that investigates sleep quality, sleep depth, restoration associated with sleep, difficulty falling asleep and trouble staying asleep, in the last week [44]. Participants respond on a fivepoint Likert scale and the total score can range from 8 to 40. Higher scores indicate more disturbed sleep. Reliability of the PROMIS-SD in the present sample was good (Cronbach alpha = 0.84; McDonald's omega = 0.88).

Analysis Strategy
First, we ran a CFA with Robust estimator to investigate the latent factor structure against alternative models. In particular, fit statistics (χ 2 , comparative fit index (CFI), Tucker-Lewis Index (TLI), root mean square error of approximation (RMSEA)) for the four-factor model were compared to those of a one-factor model. Hu and Bentler [45] suggest that values of RMSEA and standardized mean square residual (SRMR) ≤0.08 (better if ≤0.05) and CFI and non-normed fit index (NNFI) values above ≥0.95 are indicative of good fit. The threshold for good construct reliability for Cronbach alpha and McDonald's omega was set at 0.70 or above.
Second, we investigated the invariance of factor structure across gender using a multiple-group structural equation model (MG-SEM). A series of tests comparing fit statistics of hierarchical models was considered in order to investigate: configural invariance (same 4-factor model fitted on the two sub-samples with factor loadings and error variances free to vary across samples) and metric invariance (the configural model plus invariant factor loadings across the two samples and error variances free to vary).
Third, we investigated the relationships of the four TDM factors with clinical personality factors (Negative Affect, Detachment, Antagonism, Disinhibition and Psychoticism) and Sleep Disorder. In particular, we were interested in verifying whether the PID factors represented risk or protection factors in the relationship between dependence on mobile phone and sleep disturbance. In this view, we tested several moderation models considering each of the PID factors as a moderator, one at a time.

Factorial Invariance of the TDM-Brief Four-Factor Structure across Gender
In order to investigate the hypothesis that the same four-factor structure is invariant across gender sub-groups, we ran MG-SEM with hierarchical invariance tests. As can be seen from the comparison of robust fit statistics for invariance tests shown in Table 2, the fit of the four-factor TDM-brief-Ita model with invariance of factor loadings across gender (metric invariance) was not significantly different from the configural model (Satorra-Bentler robust ∆χ 2 (8) = 11.88, p = 0.157). Also, the model with invariance of intercepts (investigating the invariance of item means across gender) showed a fit not statistically different from that of the invariance model (Satorra-Bentler robust ∆χ 2 (8) = 4.24, p = 0.835). Finally, the test for invariance of latent means across gender (that corresponds to a multivariate analysis of variance (MANOVA) on the latent means across gender) was also not significant (Satorra-Bentler robust ∆χ 2 (4) = 7.47, p = 0.113) meaning that each of the four factors have the same latent mean across gender. Thus, no differences emerged between the factor structure across gender. Note: Df = degree of freeedom; AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; χ2 = Chi-squared statistic; ∆χ2 = difference in Chi-squared statistic between two hierarchical models; ∆ Df = difference in degree of freedom between two hierarchical models; Pr (∆χ2) = p-value of the ∆χ2. Table 3 presents the descriptive statistics for the four TDM-brief-Ita factors, personality factors and sleep disturbance for the total sample and as a function of gender. Comparing personality factors and sleep disturbance means across gender, we found significant differences for Negative Affect (F(1, 526) Table 4 outlines the Spearman correlations, corrected for family-wise error rate, among TDM-brief-Ita factors, personality factors and sleep disturbance. All four TDM factors correlated positively and significantly with personality factors as well as with sleep disturbance.

Moderation Effect of the Personality Inventory for Diagnostic and Statistical Manual of Mental Disorders (DSM-5) Brief Form (PID-5-BF) Factors on the Relationship between TDM-Brief Factors and Sleep Disturbance
To investigate whether PID factors play a moderator role in the relationship between TDM total score and sleep disturbance, we performed a moderation model considering each PID factor one at a time. Table 5 shows the results for the five moderation models. Only two of the five moderation models reported a significant interaction effect between TDM total score and a specific PID factor. In particular, we found that Antagonism significantly increased sleep disturbance (b = 1.214, s.e. = 0.268, p < 0.01) that significantly moderated the effect of TDM total score on sleep disturbance (b = −0.579, s.e. = 0.243, p = 0.017). Simple slopes analysis showed that for low levels of Antagonism (16 • percentile), the effect of TDM total score on sleep disturbance was positive and significant (b = 1.797, s.e. = 0.376, p < 0.001) while for high levels of Antagonism, the relationship was still positive but not significant (b = 0.651, s.e. = 0.351, p = 0.064). Therefore, it seems that low levels of Antagonism may constitute a risk factor and increase the effect of TDM score (i.e., levels of PMPU) on sleep disturbance. Likewise, we found that Psychoticism significantly increased sleep disturbance (b = 1.096, s.e. = 0.275, p < 0.01) and significantly moderated the effect between TDM total score and sleep disturbance (b = 0.503, s.e. = 0.246, p = 0.041). However, in this case, simple slopes analysis indicated that the effect of the moderator was in the opposite direction. In particular, for low values of Psychoticism

Discussion
The conceptualization and definition of PMPU as a behavioral addiction is a complex aspect under debate between researchers. The diagnosis of mobile phone addiction still remains controversial: there are no widely shared diagnostic criteria and moreover, it does not appear in any official diagnostic system, including the DSM-5. The use of multiple terms for referring to this condition concurs in creating further confusion for its evaluation (e.g., Problematic Mobile Phone Use, Mobile-Phone Addiction, Mobile-Phone Dependence, Problematic Smartphone Use, Smartphone Addiction). The same lack of clarity characterizes self-report questionnaires measuring PMPU, as tools developed from multiple definitions and different diagnostic criteria. Therefore, data are limited by methodological difficulties affecting both the diagnosis and the heterogeneity of diagnostic tools [12,14,46]. Considering the wide range of negative outcomes associated to the problematic use of mobile phones [19,[23][24][25][26][27][28][29][30]47], validated and reliable tests for its investigation are needed.
Regarding the Italian version of the TMD-brief, our findings demonstrated good reliability and significant validity, supporting the measure as a sensitive and appropriate instrument for the assessment of the problematic use of mobile phones in adolescence. First, CFA pointed out that a four-factor model showed a better fit than the unique factor solution and was invariant across gender sub-groups. The four factors were Abstinence, Abuse and interference with other activities, Tolerance and Lack of control, confirming the four-factor structure found in the original study by Choliz et al. [12]. In the present study, these four factors positively correlated with both pathological personality traits and sleep disturbances. Therefore, the TMD-brief-Ita captures the core elements of addiction to mobile phones [13]. While several instruments exploring PMPU have been validated [48], the TMD-brief has the added value of having been originally validated in participants from seven countries representing three different continents (i.e., America, Asia, Europe) [12]. The TMD-brief and other validated scales used to examine PMPU have established criterion validity using the criteria for gambling or substance-use disorders [35]. Indeed, the factors obtained reflect behavioral addiction-related concepts such as tolerance, withdrawal, craving and negative life consequences.
Second, considering the fact that the use of mobile phones becomes problematic or addictive when it leads to negative consequences [46,49], our study found that PMPU interfered with daily activities and specifically, predicted sleep disturbances. This result is consistent with previous studies among adolescent populations [50][51][52]. A possible explanation is that mobile phone use during the evening can cause a physiological activation that makes relaxation difficult and deteriorates sleep quality [53]. Previous studies have shown the negative effect of evening exposure to computers or mobile phones' screen light in suppressing the production of melatonin and consequently, affecting the circadian rhythm [54,55]. This result requires further attention considering the complex relationship between sleep, problematic technology use and the psychophysical health of adolescents [50,56].
Moreover, findings of the present study indicated that Antagonism and Psychoticism personality traits moderated the relationship between PMPU and sleep disturbances. Antagonism refers to behaviors that put one at odds with other people, including ideas of self-importance and callousness towards and/or unawareness of others. Results showed that low self-reported levels of Antagonism may constitute a risk factor and increase the effect of PMPU on sleep disturbances. We assume that PMPU in adolescents with low levels of antagonism (i.e., high levels of agreeableness) lead to increased prosocial online behaviors thereby disturbing sleep [57,58].
Results of moderation analyses also highlighted that high levels of Psychoticism increased the effect of PMPU on sleep disturbances. Psychoticism refers to the tendency to exhibit culturally incongruent, odd, eccentric, or unusual behaviors and cognitions, both in content and process. Considering the relationship between Psychoticism and emotion dysregulation [59], adolescents with PMPU and high levels of Psychoticism could experience an increase in sleep disturbances as a consequence of the difficulty in regulating changes in emotional experience elicited online.
Finally, Neuroticism (i.e., frequent and intense experiences of a wide range of negative emotions), Detachment (i.e., avoidance of socioemotional experience, including both withdrawal from interpersonal interactions and restricted affective experience and expression), and Disinhibition (i.e., orientation towards immediate gratification and impulsivity with no regard for past learning or future consequences) personality traits did not moderate the relationship between PMPU and sleep disturbance. These results are in line with those of a previous study reporting that PMPU significantly predicted sleep quality regardless of the level of a self-regulation trait [60].
Although different instruments for PMPU have been used [16] and adapted [61][62][63] in Italian populations of adolescents and adults, the present study on the adaptation of the TMD-brief to the Italian context has the advantage of primarily focusing attention on the adolescent population. De Pasquale et al. [61] reported the Italian validation of a 10-item questionnaire among 633 adolescents and young adults showing a one-factor structure. However, they did not examine the relationship between scores on the questionnaire and other variables. Pancani et al. [62] developed the Smartphone Impact Scale, a self-report scale to explore the cognitive, affective, social and behavioral impact of smartphone use among Italian adults. The analyses showed a seven-factor structure for the questionnaire, namely: Loss of control of smartphone use, Nomophobia, Smartphone-mediated communication, Emotion regulation through smartphone usage, Smartphone support to romantic relationships, Smartphone tasks support and Awareness of smartphone negative impact. Soraci et al. [63] tested the psychometric properties of a previous 6-item questionnaire among 205 Italian adults showing a unidimensional factor. By contrast, we specifically focused on the problematic use of smartphones among Italian adolescents. Importantly, our findings highlight the associations between PMPU as explored by the TMD-brief-Ita and measures of clinical personality traits and sleep disturbances. Therefore, we recommend the use of the TMD-brief-Ita as a screening tool to investigate the problematic use of mobile phones by Italian adolescents in both empirical research and clinical practice [31]. Findings from the present study indicate that when faced with adolescents with PMPU, clinicians should investigate both possible risk factors and negative repercussions. We showed that certain adolescents are more susceptible to the effects of PMPU on sleep disturbances. This finding may also be used to plan prevention interventions designed to improve healthy habits related to sleep and smartphone use while avoiding potential negative repercussions on school functioning.
Future research should pay attention to the differentiation of the normal and problematic use of smartphones. As pointed out by a previous study [48], a standard cut-off point establishing the PMPU has yet to be established. Furthermore, research needs to disentangle a complex issue, that is, whether the smartphone is a means to satisfy other addictions or if it is the object of the addiction itself, regardless of the applications and activities that are carried out through it. Finally, the understanding of the cause-effect relationship between PMPU and psychophysical health requires additional research efforts.
The present study has some limitations. First, the cross-sectional study design does not allow causal inferences on the relationships between the variables of interest. Second, data were collected using self-report measures, which may produce response biases due to social desirability. Third, we did not explore the specificities of PMPU (e.g., dysfunctional mobile phone use may often be related to interpersonal processes) or the degree of use of various mobile phone applications (such as gaming, gambling, social networking and shopping applications, etc.). Future studies should investigate these aspects and their differential influences on negative consequences [11]. Fourth, the sample was composed of secondary school students. Therefore, generalization of our results to other groups or populations should be made with caution.

Conclusions
The findings of the present study demonstrate that the Italian version of the TMD-brief demonstrated good psychometric properties to examine the problematic use of mobile phones in Italian adolescents. The four main factors, Abstinence, Abuse and interference with other activities, Tolerance and Lack of control, correlated with clinical personality traits and sleep disturbances. The moderation analyses indicated that PMPU predicted sleep disturbances, and that levels of Antagonism and Psychoticism influenced this relationship. Table A1. Italian version of TDM-brief [12]. Note. The scale for items 4, 5, and 6 ranges from 0 (Never) to 4 (Frequently); while for all other items the response scale ranges from 0 (Completely disagree) to 4 (Completely agree); Nr = item number; F = factor; Abs = Abstinence; Abu = Abuse; Tol = Tolerance; LoC = Lack of Control.