Hikikomori Risk Inventory (HRI-24): A new instrument for evaluating Hikikomori in both Eastern and Western countries

Abstract

Background:

Hikikomori, a clinical condition widely studied in Japan, is receiving increasing attention in Western Countries.

Aims:

This study aimed to create a new instrument for evaluating the risk of Hikikomori in both Eastern and Western countries.

Methods:

Through two studies on Italian adolescents, youths, and adults (n = 1,285) and a study on Japanese youths and adults (n = 392), we analyzed the psychometric properties of the Hikikomori Risk Inventory (HRI-24).

Results:

We found support, in both the Italian and Japanese samples, for the good and stable factor structure of the scale (CFI = 0.94, RMSEA = 0.057 for both the adult samples), as well as for its convergent and divergent validity.

Conclusions:

The HRI-24 might be used in future studies in both Western and Eastern countries to shed light on the features of this clinical disorder in different cultures. This will allow the development of culture-sensitive preventive and clinical interventions.

Keywords

Adolescents Hikikomori social isolation social withdrawal youths

Introduction

Hikikomori, or extreme social withdrawal, is a psychosocial disorder that has received much attention since the 90s in Japan (Funakoshi & Miyamoto, 2015). In line with this, the Hikikomori phenomenon has often been considered as a condition specific to the Japanese context, namely as a cultural-bound syndrome (Aguglia et al., 2010; Teo & Gaw, 2010). However, recently, it is becoming more and more common in other countries, including Western ones (Malagón-Amor et al., 2015). Cases were reported in the United States (Teo, 2010, 2012), Oman (Sakamoto et al., 2005), South Korea (Lee et al., 2013), Hong Kong (Chan & Lo, 2014), India (Teo et al., 2015), France (Guedj-Bourdiau, 2011; Maia et al., 2014), and Italy (Ranieri, 2015, 2016; Ranieri et al., 2015). Besides these case reports, some surveys showed that Hikikomori exists in other countries besides Japan, especially in urban areas, such as Australia, Bangladesh, Iran, Thailand, and Taiwan (Kato et al., 2012). Hence, Hikikomori seems to be a transcultural condition (De Michele et al., 2013; Sakamoto et al., 2005; Stip et al., 2016; Watts, 2002).

However, currently, there is no consensus about Hikikomori’s conceptualization as a syndrome or as a specific cultural condition (Tajan, 2015; Teo & Gaw, 2010). Non-Japanese Hikikomori cases may have some peculiar features related to each specific country’s culture (Sarchione et al., 2014). The exponential growth of Hikikomori in Western countries could be related to “the fragmentation of social structures in the late modern period” (Furlong, 2008, p. 309), and to cultural and social dimensions such as individualism, high youth unemployment rates, traditional value crisis, and the development of new virtual communication technologies, which are typical Western features (Sarchione et al., 2014). Thus, it is not surprising to see that Hikikomori is becoming a trans-cultural phenomenon in the post-modern era.

Beyond cultural-specific differences, Hikikomori may be defined as a psychopathological and social syndrome that can occur from early adolescence. Its onset is typically around 13 years (Malagón-Amor et al., 2015) and is commonly spread among people between 15 and 39 years (Tajan et al., 2017), although there are Hikikomori cases over the age of 40 (Umeda & Kawakami, 2012) and the age is increasing according to recent research from the Japanese Cabinet Office (2019). Hikikomori’s three main characteristics are complete withdrawal from society for at least 6 months, lack of interest or willingness to attend school/work, and lack of interest in personal relationships (Teo, 2010). Saitō (1998) suggested social withdrawal as the primary symptom, manifesting itself in various ways and degrees. The confinement in owns’ rooms may last for a few months or up to several years (Aguglia et al., 2010). Moreover, it should be taken into account that, in students’ case, usually, the first manifestation of social withdrawal is school phobia (Aguglia et al., 2010).

In order for withdrawn people to be defined as Hikikomori, they should not have schizophrenia, mental retardation, or other mental disorders that could explain their social withdrawn behavior (Kaneko, 2006; Kato et al., 2012; Saitō, 1998; Tajan, 2015; Tajan et al., 2017; Tateno et al., 2012; Teo, 2010; Teo & Gaw, 2010; Watts, 2002). However, besides differential diagnosis, it should be noted that people with Hikikomori may have comorbidity with psychotic disorders, including schizophrenia (Tajan, 2015). Also, they may have comorbidity with other clinical diagnoses, such as affective-disorders, anxiety disorder, obsessive-compulsive disorder, personality disorder, or pervasive developmental disorder (Kondo et al., 2007, 2013; Koyama et al., 2010; Li & Wong, 2015; Suwa et al., 2003).

Among additional primary diagnoses associated with Hikikomori, there are various types of phobias. Social phobia and specific phobia preceding the onset of Hikikomori have been reported in 14.8% and 12.5% of Japan’s cases in a World Mental Health Japan survey between 2002 and 2006 (Koyama et al., 2010). Hikikomori people can have anthropophobia, or Taijin Kyofusho, namely the fear of people and social contacts, although studies distinguish Hikikomori from Taijin Kyofusho as different conditions (Suwa & Suzuki, 2013). This phobia usually develops secondarily to social withdrawal and, if present, might lead to a worsening of the clinical condition. They may also have agoraphobia, namely the avoidance of places where it would be difficult to get assistance in case of a panic attack or a high state of anxiety (De Luca, 2017). Sometimes, there could be automisophobia (the fear of getting dirty), which can turn into personal hygiene-related obsessive-compulsive symptoms. They can wash their bodies or hands so often a day for fear of being contaminated by dirt that their hands are so worn as to appear without skin (Saitō, 1998, 2010).

Besides phobia, there are other typical symptoms. They usually have depressed mood and depressive symptoms, such as recurrent thoughts about death, suicidal ideation, lethargy, apathy, feelings of guilt, and worthlessness (Saitō, 2010). Sometimes, there are also regressive behaviors, like the inability to manage anger, which might be associated with violent behaviors toward people or objects (Saitō, 1998). In severe cases, there may also be a loss of contact with reality, excessive suspiciousness, and the manifestation of persecutory ideation concerning both the outside and family members (Aguglia et al., 2010; Saitō, 2010).

Another typical Hikikomori feature is the alteration of the sleep-wake rhythm: they sleep many hours during the day and stay awake at night for playing video games, reading manga, or surfing on the web (Aguglia et al., 2010). Hence, many people with Hikikomori use Internet profusely and can spend more than 12 hours a day using a computer (Stip et al., 2016). For this reason, some scholars proposed comorbidity with Internet addiction (De Michele et al., 2013; Kato et al., 2012; Lee et al., 2013). Lee et al. (2013) reported that up to 56% of the South Korean Hikikomori cases of their sample were at-risk for Internet addiction, while 9% had a diagnosis of Internet addiction. However, in contrast with people with a diagnosis of Internet addiction, using the Internet could improve the Hikikomori’s quality of life since it provides them with a way to meet people with similar interests and problems (Taylor, 2006). Hence, the use of the Internet could be considered as a positive factor and not as a comorbid diagnosis (Stip et al., 2016), given that many Hikikomori use the Internet in an adaptive way for keeping in touch with the world outside their rooms (Teo et al., 2015). However, we should note that new technologies could promote social withdrawal (Kato et al., 2017) and be aware of the difference between adaptively using the Internet and being addicted to the Internet.

Finally, about Hikikomori risk factors, some studies reported a correlation between Hikikomori and childhood traumatic experiences (Hattori, 2006; Kaneko, 2006; Stip et al., 2016; Teo, 2010). Moreover, many Hikikomori have been bullied at school or suffered from other forms of peer rejection (Funakoshi & Miyamoto, 2015; Lee et al., 2013; Maia et al., 2014; Teo, 2010). Most of them also have an insecure ambivalent attachment style and a shy-inhibited temperament (Krieg & Dickie, 2013; Teo & Gaw, 2010). Other Hikikomori risk factors include parental rejection (Krieg & Dickie, 2013), overprotection within the family context (Li & Wong, 2015), caregivers’ severe psychopathology (Umeda & Kawakami, 2012), and dysfunctional family relationships (Chan & Lo, 2014; Lee et al., 2013).

The present study

Given the recent widespread of Hikikomori in Western countries, it is vital to have a self-report instrument that is brief and easy to administer in the general population, from adolescence (when Hikikomori usually arises; Malagón-Amor et al., 2015) to adulthood (where some cases might still be present; Umeda & Kawakami, 2012). This scale might be used for research aimed at detecting typical Western Hikikomori features, as well as the main adverse consequences associated with this condition, in order to develop culture-sensitive preventive and clinical interventions. Hence, having an instrument validated both in Eastern and Western countries is imperative to compare the results from a cross-cultural perspective.

In East Asia, the typical Hikikomori profile is a young male (often the eldest son) living with a middle-class family who shows dependency on his parents (especially the mother; Saitō, 1998). The typical Hikikomori in the West is also a young male with relatively high education and usually living with his family (Sarchione et al., 2014), and the family typically has disrupted dynamics (Malagón-Amor et al., 2015). However, in contrary to cases in Japan, where 45.5% of Hikikomori cases have not been diagnosed with another psychiatric disorder (known as Primary Hikikomori; Koyama et al., 2010), in Spain (i.e., a Western country), only 1.8% of cases do not have comorbid psychiatric diagnoses (Malagón-Amor et al., 2015).

This difference may be explained by the different risk factors of Hikikomori across cultures. In East Asian societies such as Japan, the risk of becoming Hikikomori is more related to social factors, such as bullying at school (Saitō, 1998), failure in education (Uchida, 2010) or employment (Furlong, 2008). These negative experiences impede them from achieving the cultural goals for being accepted in a collectivist society. As a result, they lose the desire and motivation to participate and choose to retreat from society itself (Toivonen et al., 2011). In Western societies, where individual characters are valued, the societal pressure to conform and to be accepted is not as strong as in Eastern cultures. As a result, the risk of becoming Hikikomori is more related to individual personality and psychological factors (Ovejero et al., 2014), such as depression (Teo, 2012), introversion and paranoia (García-Campayo et al., 2007), and social withdrawal serves as a coping strategy for anxiety or nervous breakdowns (Teo et al., 2015).

To the best of our knowledge, in literature, there are only two self-report instruments that have been proposed for evaluating Hikikomori so far. The first one is the NEET-Hikikomori Risk Factors Scale (NHR; Uchida & Norasakkunkit, 2015). This scale considers Hikikomori as part of a spectrum of psychological tendencies linked to the risk of being marginalized in the society and distinguishes Freeter, NEET (Not in Employment, Education, or Training) youths (Genda, 2005), and Hikikomori on different levels of marginalization risks (Smith, 2015; Uchida & Norasakkunkit, 2015). As pointed out by Teo et al. (2018), this scale is mainly based on occupation withdrawal. The other scale is the 25-item Hikikomori Questionnaire (HQ-25; Teo et al., 2018), which focuses on social withdrawal; the scales are indeed Socialization, Isolation, and Emotional Support.

Hence, these scales do not evaluate other psychopathological typical features associated with the risk of Hikikomori. Finally, both the NHR and the HQ-25 have been validated on East Asians only, and they have not been evaluated through Confirmatory Factor Analyses, which is critical to support the stability of the factor solutions found through Exploratory Factor Analyses.

Hence, we conclude that the current literature still lacks a self-report instrument specifically developed to screen for the risk of the Hikikomori condition in the Western population, which is also valid in Eastern countries. Hence, this study aims to address this gap by creating and analyzing the psychometric properties of a new measure for the screening of Hikikomori in the general population, the Hikikomori Risk Inventory (HRI-24). The HRI-24, measuring the symptoms usually present in Hikikomori (such as depressive symptoms and phobias), could be a useful instrument to assess the risk of this psychosocial condition in both the Eastern (i.e., where Hikikomori originated) and the Western (i.e., where Hikikomori is now growing) countries. More specifically, we will evaluate the psychometric properties of our scale on both Italian and Japanese participants.

In order to reach our aim, we performed three studies: (i) we created a pool of 79 items addressing Hikikomori symptomatology to be reduced through Exploratory and Confirmatory Factor Analyses on a sample of Italian adolescents, youths, and adults; (ii) once we reached the final 24-item version, we deepened the HRI-24 psychometric analyses on another Italian sample. More specifically, we repeated Confirmatory Factor Analyses for cross-validation, and we analyzed its convergent and divergent validity; (iii) finally, we evaluated if the HRI-24 holds the factor structure and good psychometric properties we found on Italian samples also on a Japanese sample, that is an Eastern country. Hence, we repeated Confirmatory Factor Analysis and convergent and divergent validity analyses.

Study 1: Development of the Hikikomori Risk Inventory and factor analysis on an Italian sample

In the first stage of our research, we created an extensive pool of 79 items referring to the Hikikomori symptomatology described by Saitō (1998) and to its main features, as reported by the literature: depressed mood, feelings of worthlessness, feelings of guilt, lethargy, thoughts of death, acting-out, agoraphobia, anthropophobia, erythrophobia, paranoia, internet addiction, reversed sleep cycle, school phobia, and bullying. Except for bullying items, the reference time is “the last 6 months.”

Next, in order to reduce the number of items to have a quick screening instrument for the general population, we performed factor analyses on an Italian sample, including adolescents (aged 13–18), youths (aged 19–30), and adults (aged 31–50).

Methods

Participants

We recruited 893 Italians (31.5% males, 68.5% females) living in Central Italy and aged between 13 and 50 years (M = 29.15; DS = 10.82). The sample comprehended students from secondary schools of second grade and colleges (36.1%), workers or people looking for a job (59.6%), and a few participants who reported to be both students and workers (4.3%). In terms of highest education, 0.2% of the participants completed primary school, 32.4% completed first grade of secondary school, 30.3% hold a high school diploma, 10.2% had a Bachelor’s Degree, 18.3% had a Master’s Degree, and 8.6% hold a Doctoral Degree or other post-graduate specializations. We did not foresee exclusion or inclusion criteria, since we aimed to create a scale that evaluates the risk for Hikikomori in the general population.

In order to perform the analyses, we randomly divided this total sample into two sub-samples. We performed Exploratory Factor Analysis (EFA) on the first sample (n = 449, 29.2% males, 70.8% females; M age = 28.63 ± 10.77) and Confirmatory Factor Analysis (CFA) on the second sample (n = 444, 33.8% males, 66.2% females; M age = 29.68 ± 10.85).

Measures

Hikikomori Risk Inventory: Pilot version

The Italian authors of this study developed an initial pool of 79 items referring to the symptomatology described by Saitō (1998) and to the Hikikomori main features reported by the literature. In this stage, we administered this 79-item pilot version. The subjects have to answer each item through a 5-point Likert scale ranging from 1 (Strongly disagree) to 5 (Strongly agree).

Procedure

As a first step, we obtained the authorization to conduct the study by the Ethical Committee of the Department of Health Sciences of the University of Florence. Moreover, before administering the instruments, we gathered the Informed Consent from the participants. For participants under 18, we asked for the Informed Consent of their parents too, and they filled a paper-and-pencil questionnaire. Participants over the age of 17 filled an online questionnaire. On the first page of the questionnaire, we presented the research aims and all the information required in the Informed Consent. We specified that by filling in the questionnaire on the next pages, the participants gave us their informed consent to participate in the research. All participants completed the 79-item version of the Hikikomori Risk Inventory (HRI) and answered demographic questions asking about age, gender, level of education, and professional activity.

Data analysis

We performed data analyses using SPSS.25 and AMOS.20. We randomly divided the total sample into two subsamples. On the first subsample (n = 449), we conducted some EFAs (Principal Axis Factoring, Promax rotation) to reduce the 79-item pilot version of the HRI and highlight its factor structure. Then, on the second sample (n = 444), we carried out a CFA (Maximum Likelihood estimation) for analyzing the fit of the EFA solution to the data. In order to evaluate the model’s goodness of fit, we referred to the Goodness of Fit Index (GFI), the Tucker-Lewis index (TLI), and the Comparative Fit Index (CFI). Indeed, while the χ² index is influenced by the sample size (Bentler & Bonnet, 1980), the GFI, TLI, and CFI are relative indexes, and hence they apply to both large and small samples. For these indexes, values greater than or equal to 0.90 are considered satisfactory (Schermelleh-Engel et al., 2003). Moreover, we referred to the Root Mean Square Error of Approximation (RMSEA), for which these are the cut-off values: values below 0.05 indicate an excellent fit; values between 0.05 and 0.08 indicate an acceptable fit (Reeve et al., 2007). Finally, we calculated the HRI-24 scales’ reliability on the total sample (n = 893) through Cronbach’s alpha coefficient.

Results

First, we conducted some EFAs on the first sample (n = 449) to reduce the 79-item pilot version of the HRI and highlight its factor structure. We found the best factor solution for a 24-item and 5-factor version (HRI-24). We selected the five factors based on both the scree-plot and the criterion of the eigenvalues greater than one. The eigenvalues are 11.25, 3.42, 2.52, 1.65, 1.43 (with the sixth factor having an eigenvalue of 0.32). Moreover, these five factors account for 84.47% of the total variance. The first factor accounts for 46.86% of the variance, while the cumulative explained variance from factor 2 to factor 4 is, respectively: 61.13%, 71.64%, 78.51%. The values of communalities are very high for all the items, ranging between 0.71 and 0.89. Moreover, all the saturation values are higher than 0.80 (see Table 1). Referring to the content of the items saturating on the five factors, we labeled them: (1) Paranoia, (2) Depressive Mood, (3) Lethargy, (4) Anthropophobia, and (5) Agoraphobia. Hence, we calculated the internal reliability of these scales, which showed values higher than 0.93 (see Table 1).

Table 1.

Hikikomori Risk Inventory (HRI-24). Exploratory factor analysis, 24-item and 5-factor, n = 449.

Hikikomori Risk Inventory item	Para	Dep	Let	Ant	Ago
1. I feel severely anxious when I meet people outside				0.88
2. I am uncomfortable when I am in touch with others				0.91
3. I have great difficulty in participating in social activities				0.88
4. I am afraid of social relationships				0.85
5. I feel very anxious when I am in a closed space (e.g., shops, cinema, theater, bank, etc.)					0.96
6. I am afraid of being in crowded places					0.91
7. I avoid going to public places when I go out alone					0.83
8. When I am in crowded places, I feel some strange sensations (accelerated heartbeat, excessive sweating, tremors, choking sensations, etc.)					0.88
9. I am diffident and suspicious of others	0.83
10. Confiding with others is dangerous because people might then reveal my secrets	0.93
11. I am afraid of being deceived by others	0.87
12. I prefer doing things on my own because I do not trust people	0.87
13. People are willing to do anything to achieve their goals	0.92
14. I only trust myself	0.90
15. If others do me a favor, I do not believe in their good intention	0.91
16. I feel tired and fatigued			0.93
17. I manage to do my activities with difficulty because I feel powerless			0.93
18. I often feel weak and lacking in energy			0.94
19. I sleep many hours a day because I often feel weak			0.88
20. There are moments in which nothing seems important		0.86
21. Nothing is able to really thrill me		0.85
22. I rarely feel positive emotions		0.90
23. I feel dispirited and hopeless for the future		0.80
24. I feel a sense of inner emptiness		0.92
α	0.97	0.94	0.96	0.94	0.95

Note. Extraction method: Principal Axis Factoring; Rotation: Promax; Factor loadings below 0.30 are not presented; Para = paranoia; Dep = depressive mood; Let = lethargy; Ant = anthrophobia; Ago = agorophobia.

Next, we carried out a CFA on the second sample (n = 444) to cross-validate this 24-item and 5-factor version of the HRI-24. The fit indexes indicate that this model has a good fit: χ²/df = 1.88, p < .001; GFI = 0.92; CFI = 0.98; TLI = 0.98; RMSEA = 0.045 (0.038–0.051). Moreover, factor correlations range between 0.37 (Lethargy and Paranoia) and 0.63 (Anthropophobia and Agoraphobia), suggesting that a total score might be calculated to have a Hikikomori risk score. This total score has high internal reliability: α = 0.95. Figure 1 reports a graphical representation of the model.

Figure 1.

Five-factor model, Hikomori Risk Inventory (HRI-24), Italian sample (n = 444).

In conclusion, this first study provides support for the good and stable factor structure of the HRI-24, and the good internal reliability of the five scales. It is interesting to note that EFA analyses led to deleting the items related to Hikikomori risk factors, such as being bullied at school. The items included in the final HRI-24 are mostly related to psychological disorders that characterize Hikikomori. This might be due to the fact that in Western and Eastern countries there are different risk factors for Hikikomori (e.g., Ovejero et al., 2014; Saitō, 1998; Uchida, 2010). Hence, the final version of HRI-24 – as being focused on common characteristics in Eastern and Western countries – seems more suitable to be used across these different cultures.

Study 2: Further psychometric analysis on an Italian sample

Since the results of the previous study provided support for the good factor structure of the HRI-24, we deepened the psychometric analyses of this version on a new Italian sample, which received the reduced version. Hence, we performed a CFA for cross-validation, and we analyzed the convergent and divergent validity of the HRI-24. In this version of the HRI-24, we added a head-sheet comprehending some Hikikomori-related questions, which could help the clinician understand if the high HRI-24 total score is specifically due to extreme social withdrawal. More specifically, we asked: (i) how many hours a day he/she stays closed in his/her bedroom; (ii) how many hours a day he/she spend awake doing some activities after 11 p.m. (i.e., when people are usually going to sleep or are already sleeping); (iii) if he/she is interested in social relationships (yes/no option); (iv) if he/she prefers to communicate through the virtual world as compared to the real world (yes/no option); (v) if in the last 6 months he/she has reduced or gave up one or more daily activities (yes/no option), which was followed by the request to specify the reduced/abandoned activity in case of “yes” answer.

Methods

Participants

We recruited 392 Italian youths and adults (77% of females) living in Italy. Most of them were of Italian origin (96.4%) and lived in Tuscany (73.2%). All the participants could read and understand Italian. The participants were aged between 18 and 64 years (M = 38.02; DS = 10.51). However, participants over 50 years constitute only 12.2% of the sample, while participants over 60 years are 1.4%. In terms of highest education, 0.5% of the participants completed primary school, 17.6% completed first grade of secondary school, 44.1% hold a high school diploma, 11.5% had a Bachelor’s Degree, 20.2% had a Master’s Degree, and finally 6.1% hold a Doctoral Degree or other post-graduate specializations. Like for Study 1, we did not foresee exclusion or inclusion criteria.

Concerning Hikikomori behaviors, the participants reported staying awake after 11 p.m. between 0 and 11 hours (M = 1.15; DS = 1.30) and staying closed in their bedroom between 0 and 20 hours (M = 2.99; DS = 3.55) per day. However, some of the participants answering with a number ranging between 5 and 9 specified that these were their typical hours of sleep. Hence, we speculated that some Italian participants, who are probably not familiar with Hikikomori and the related behavior of “being closed in the bedroom,” might have misunderstood the question. For this reason, we decided to change (for the final version of the scale) the related open question in order to specify that we are interested in the number of hours spent closed in the bedroom, without counting the time spent sleeping. For this study, we filtered subjects who answered within the range 0 to 4 or 10 to 20, and we found that 260 subjects (up to 392) reported these answers (M = 1.24; DS = 2.64). Almost all of these participants were in the “under 5 hours” range (97.3%).

Measures

Hikikomori Risk Inventory (HRI-24)

In this study, we administered the 24-item and 5-factor version established in the previous study, which has been preceded by a few questions that address some typical Hikikomori behaviors (e.g., staying closed in the bedroom, or staying awake after 11 p.m.). The five scales are Anthropophobia, Agoraphobia, Paranoia, Lethargy, and Depressive Mood. It is also possible to calculate a total score in order to have a Hikikomori Risk score. Participants have to answer through a 5-point Likert scale that ranges from 1 (Strongly disagree) to 5 (Strongly agree).

Brief Sensation Seeking Scale (BSSS; Hoyle et al., 2002)

In order to evaluate the divergent validity of the HRI-24, we administered the BSSS. It is a scale that evaluates sensation seeking, which is an opposite behavior compared to social withdrawal, which characterizes Hikikomori. In fact, sensation seeking is mostly related to externalizing disorders, such as substance-related addictions (Adams et al., 2012; Giannini & Loscalzo, 2020; Pokhrel et al., 2010; Stautz & Cooper, 2013). Moreover, it is associated with extraversion (e.g., Aluja et al., 2003; Zuckerman et al., 1978). The BSSS is the brief version of the form V of Zuckerman’s Sensation Seeking Scale (SSS-V; Zuckerman et al., 1978) composed of 40 forced-choice items. The BSSS, instead, is made up of eight items, two for each factor of the SSS-V: Experience Seeking, Boredom Susceptibility, Thrill and Adventure Seeking, and Disinhibition. Moreover, the BSSS response format is a 5-point Likert scale ranging from 1 (Completely Disagree) to 5 (Completely Agree). In this study, we used the Italian version of Primi et al. (2011), which showed good psychometric properties on Italian adolescents. However, since our sample is made by participants older than 17, we performed a preliminary CFA on a random subsample of 100 participants (i.e., a little more than 10 subjects for each item). The results showed a satisfactory fit to the data: χ²/df = 2.04, p = .004; CFI = 0.86; GFI = 0.91; RMSEA = 0.103 (0.056–0.147), which improved (like in Primi et al., 2011’s study) by allowing the correlation between two errors (i.e., items 5 and 6): χ²/df = 1.55, p = .059; CFI = 0.93; GFI = 0.93; RMSEA = 0.075 (0.000–0.125). For this solution, the factor loadings range between 0.37 and 0.72. Finally, in our sample, the alpha is good, as being 0.75.

NEET-Hikikomori Risk Factors Scale (NHR; Uchida & Norasakkunkit, 2015)

This scale has been developed from a pool of 53 items, and the final 27-item version has been reached through EFA. The NHR is a self-report scale that evaluates NEET and Hikikomori as a spectrum of psychological tendencies associated with the risk of being marginalized in society. Participants have to answer each item using a 7-point Likert scale ranging between “Completely Disagree” to “Completely Agree.” The Japanese NHR has three factors: Freeter Lifestyle Preference (or the conscious choice of not engaging in full-time employment despite job availability; 14 items, 6 of which are reverse items; α = 0.83); Lack of Self Competence (11 items, 3 of which are reverse items; α = 0.83); Unclear Ambitions for the Future (two items; no reversed item; α = 0.79). Uchida and Norasakkunkit (2015) also suggest using an overall score (α = 0.82), since in the Japanese context the combination of all three factors increases the chances of becoming NEET/Hikikomori.

Since the Italian version of this scale is not available, after obtaining permission to validate the scale from the authors, we performed some preliminary factor analyses on our Italian translation. First, we checked the goodness of the Italian translation, which has been reached using the back-translation process. Next, we conducted both EFA and CFA analyses. Uchida and Norasakkunkit (2015) did not conduct CFA to cross-validate the factor structure of the NHR; moreover, they selected three factors (eigenvalues: 6.33, 4.31, 2.89) against evidence of other eigenvalues higher than 1 (i.e. 2.07 and 1.86), based on the consideration that “the loading score had decreased by the third factor” (p. 1118). For these reasons, given the uncertain factor stability of the scale, as a first step, we conducted EFAs on the total sample, in order to have more than ten subjects (an adequate sample size) for each NHR item.

The EFA (Principal Factor Analysis, Promax rotation) that we conducted on our total sample showed that the original version of the scale does not have a clear factor structure on the Italian sample. Hence, deleting the items whose communality was very low, as well as the items loading on two or more factors, we reached a satisfactory solution for a 16-item and 4-factor version: (i) Social Issues (items 15, 17, 18, 23 – saturation values between 0.34 and 0.82, α = 0.77); (ii) Lack of Self Competence (items 19, 20, 21, 24 – saturation values ranging between 0.33 and 0.87, α = 0.80); (iii) Freeter Lifestyle Preference (items 1, 2, 4, 6, 8, 10 – saturation values ranging between 0.35 and 0.64, α = 0.65); (iv) Unclear Ambition for the Future (items 26, 27 – saturation values of 0.86, α = 0.84).

Next, in order to cross-validate this NHR-16 Italian version, we conducted two CFAs on a random half of our sample (n = 184). We compared the original 27-item and 3-factor version with the 16-item and 4-factor solution we found by means of our EFA. The results highlighted a better fit for our Italian shorter version. More specifically, the original version showed the following fit indexes: χ²/df = 2.167, p < .001; CFI = 0.70; GFI = 0.75; RMSEA = 0.080 (0.072–0.088). Furthermore, all the Freeter Lifestyle Preference items loaded negatively or with standardized values near zero on the factor, besides being statistically non-significant. The 16-item and 4-factor version, instead, reported statistically significant and positive standardized factor loadings for all the items, and a good fit to the data: χ²/df = 1.56, p < .001; CFI = 0.94; GFI = 0.91; RMSEA = 0.055 (0.038–0.072). Hence, we decided to refer to this version for our subsequent convergent validity analyses. However, future studies should further evaluate this scale on Italian populations to understand if the different factor structure is due to different Western and Eastern cultural backgrounds. The most problematic scale is indeed the Freeter Lifestyle Preference (which also has a low alpha value in our reduced version). This could be explained by the fact that in Italy, currently, there is a lack of full-time permanent jobs; as a consequence, the items of this scale might not apply to the Italian context, as Freeter Lifestyle Preference would not be perceived as a realistic option in the current economic situation.

Procedure

For this study, we created an online questionnaire. On the first page of the questionnaire, we presented the aims of the research and all the information required in the Informed Consent. We specified that by filling in the questionnaire on the next pages, the participants gave us their informed consent to take part in the research. The questionnaire asked participants to provide personal data for demographic questions (e.g., age, gender) and answer the HRI-24, the BSSS, and the NHR items.

Data analysis

We performed analyses using SPSS.25 and AMOS.20. First, we performed a CFA (Maximum Likelihood estimation) for cross-validating the factor structure found through Study 1 on a new Italian sample. Then, we calculated the internal reliability of the HRI-24 scales and of the total score again. Next, we analyzed HRI-24 convergent and divergent validity using Pearson correlations between the HRI-24 scales, the BSSS total score, and the NHR-16 scales, as well as with some indicators of Hikikomori tendency gathered through the HRI-24 head-sheet (e.g., time spent closed in the bedroom) on a random subsample of participants, namely about 25% of the total sample (n = 111).

Finally, we analyzed through seven ANOVAs group differences between those who score high (75° percentile) versus low (25° percentile) on the HRI-24 total score (the dependent variables are: age, time spent closed in the bedroom, and staying awake after 11 p.m.), as well as group differences on the HRI-24 total score (independent variables: gender, interest in social relationships, preference for virtual relationships rather than real ones, and having reduced/abandoned some activities in the last 6 months).

Results

First, we analyzed the fit of the HRI-24 on the total sample (n = 392), and we found further support for this model: χ²/df = 2.27, p < .001; GFI = 0.89; CFI = 0.94; TLI = 0.93; RMSEA = 0.057 (0.051–0.063). The standardized factor loadings range between 0.40 and 0.95, while factor correlations range between 0.32 (Lethargy and Anthropophobia) and 0.66 (Agoraphobia and Anthropophobia). Figure 2 graphically shows this model. Moreover, all the scales have good internal reliability: Anthropophobia, α = 0.91, item-total correlations between 0.85 and 0.91; Agoraphobia, α = 0.84, item-total correlations between 0.76 and 0.88; Paranoia, α = 0.82, item-total correlations between 0.56 and 0.80; Lethargy, α = 0.83, item-total correlations between 0.62 and 0.92; Depressive Mood, α = 0.88, item-total correlations between 0.76 and 0.86; HRI-24 total score, α = 0.92, item-total correlations between 0.31 and 0.73.

Figure 2.

Five-factor model, Hikomori Risk Inventory (HRI-24), Italian sample, n = 392; Japanese sample, n = 392.

Next, we analyzed the convergent and divergent validity through Pearson correlation between the HRI-24 subscales and total score, the BSSS total scores, and the NHR-16 subscales (Table 2 reports the results of these analyses; n = 111). We found evidence for the convergent validity of the HRI-24 since both its total score and subscales correlate with the NHR-16 scales (except for the NHR Freeter Lifestyle Preference scale that has statistically significant correlations with the HRI-24 total score and Lethargy subscale only).

Table 2.

Convergent and divergent validity analyses of the HRI-24 (Italian sample, n = 111).

	1	2	3	4	5	6	7	8	9	10	11	12	13
1. HRI-anthropophobia	–
2. HRI-agoraphobia	0.59***	–
3. HRI-paranoia	0.48***	0.37***	–
4. HRI-lethargy	0.25**	0.24**	0.20*	–
5. HRI-depressive mood	0.51***	0.50***	0.47***	0.63***	–
6. HRI-total	0.73***	0.67***	0.75***	0.64***	0.86***	–
7. BSSS-sensation seeking	−0.15	−0.08	0.004	0.05	−0.03	−0.04	–
8. NHR-Freeter Lifestyle Preference	0.10	0.08	0.17	0.21*	0.14	0.20*	0.22*	–
9. NHR-social issues	0.67***	0.34***	0.47***	0.34***	0.57***	0.65***	−0.11	0.15	–
10. NHR-lack of self competence	0.53***	0.42***	0.39***	0.44***	0.68***	0.67***	−0.14	0.13	0.57***	–
11. NHR-unclear ambition future	0.46***	0.24**	0.31***	0.40***	0.58***	0.55***	0.06	0.28**	0.47***	0.54***	–
12. Hours/day closed in bedroom	0.16	0.05	0.11	0.18	0.25**	0.21*	−0.07	0.24**	0.24**	0.29**	0.22*	–
13. Hours wake up after 11 p.m.	0.13	0.07	0.03	0.29**	0.28**	0.22*	0.04	0.09	0.18	0.31***	0.18	0.16	–

Note. HRI = Hikikomori Risk Inventory (HRI-24); BSSS = Brief Sensation Seeking Scale; NHR = Neet-Hikikomori Risk Factors Scale, 16-item and 4-factor Italian version.

*p ⩽ .05. **p ⩽ .01. ***p ⩽ .001.

We evaluated divergent validity using the BSSS that assesses sensation seeking, which is an opposite behavior compared to the social withdrawal that characterizes Hikikomori. The results showed that HRI-24 scales and total score have no statistically significant correlations with the BSSS.

Concerning the correlations with Hikikomori indicators, the results showed positive correlations between the HRI-24 total score and the hours spent closed in the bedroom and the hours spent awake after 11 p.m. performing activities such as reading or surfing on the Internet. Most of the correlations between these two variables and the HRI-24 scales are not statistically significant. However, as previously explained, Italians may not have correctly understood the meaning of the question related to the time spent closed in the bedroom, as many replied with the number of hours they usually spend sleeping.

Concerning group differences, we performed some ANOVAs in order to understand if the HRI-24 total score discriminates well between people at risk of Hikikomori (i.e., participants whose total score is above the 75° percentile, that is, between 59 and 108, n = 98) and people with a low risk of Hikikomori (i.e., participants whose total score is below the 25° percentile, namely between 24 and 36, n = 102). Our results showed that the HRI-24 total score discriminates well people at-risk of Hikikomori from others. Indeed, those who have a high HRI-24 total score have higher self-reported time spent closed in the bedroom and staying awake after 11 p.m., as compared to people with low HRI-24 scores. No age difference was observed (see Table 3).

Table 3.

ANOVAs results by high and low Hikikomori; Italian and Japanese samples.

Variable	Country	HRI total^a	M (SD)	n	F	gdl	p	η²
Age	Italy	Low	37.35 (9.91)	102	0.034	1,198	ns	0.03
		High	37.08 (11.01)	98
		Total	37.22 (10.44)	200
	Japan	Low	38.04 (7.70)	98	5.00	1,195	.026	0.03
		High	35.64 (7.39)	99
		Total	36.83 (7.62)	197
Hours/day closed	Italy	Low	1.63 (2.53)	94	22.08	1,184	<.001	0.11
		High	4.12 (4.45)	92
		Total	2.86 (3.81)	186
	Japan	Low	3.37 (4.03)	98	33.41	1,192	<.001	0.15
		High	7.91 (6.61)	96
		Total	5.62 (5.90)	194
Hours awake after 11 p.m.	Italy	Low	1.08 (0.79)	102	5.78	1,198	.017	0.03
		High	1.55 (1.81)	98
		Total	1.31 (1.41)	200
	Japan	Low	1.22 (1.14)	98	10.78	1,195	.001	0.05
		High	1.94 (1.84)	99
		Total	1.58 (1.57)	197

Italian low HRI-24 = 24 to 36; Italian high HRI-24 = 59 to 108; Japanese low HRI-24 = 26 to 53; Japanese high HRI-24 = 77 to 115.

Lastly, we performed other ANOVAs to evaluate if there is a difference in the HRI-24 total score between males and females, and between people who answered yes or no to the Hikikomori-related questions of the HRI-24. Except for gender, we found statistically significant differences for all the variables analyzed: not having interest in social relationships, preferring virtual relationships to real ones, and having reduced/abandoned activities in the last 6 months is associated with a higher HRI-24 score (see Table 4).

Table 4.

ANOVAs results by gender and Hikikomori-related variables; Italian and Japanese samples.

Variable	Country	Group	M (SD)^a	n	F	gdl	p	η²
Gender	Italy	Males	49.36 (16.15)	90	0.270	1,390	ns	0.001
		Females	48.35 (16.02)	302
		Total	48.58 (16.03)	392
	Japan	Males	66.82 (17.11)	197	2.73	1,390	ns	0.007
		Females	63.94 (17.39)	195
		Total	65.39 (17.29)	392
Interest in relationships	Italy	Yes	47.84 (15.62)	375	19.36	1,390	.001	0.05
		No	64.94 (16.74)	17
		Total	48.58 (16.03)	392
	Japan	Yes	60.47 (15.71)	255	69.30	1,390	<.001	.15
		No	74.54 (16.38)	137
		Total	65.39 (17.29)	392
Preference virtual relationship	Italy	Yes	67.04 (15.01)	26	40.63	1,390	<.001	0.09
		No	47.27 (15.30)	366
		Total	48.58 (16.03)	392
	Japan	Yes	73.91 (16.32)	106	38.65	1,390	<.001	0.09
		No	62.23 (16.58)	286
		Total	65.39 (17.29)	392
Gave up activities	Italy	Yes	56.40 (17.98)	85	27.54	1,390	<.001	0.07
		No	46.42 (14.77)	307
		Total	48.58 (16.03)	392
	Japan	Yes	71.44 (14.83)	61	9.04	1,390	.003	0.02
		No	64.27 (17.50)	331
		Total	65.39 (17.29)	392

Note. Interest in relationships = interest in social relationships; Preference virtual relationship = preference for virtual relationships as compared to real ones; Gave up activities = having reduced or give up one or more activities in the last 6 months.

The values refer to the HRI-24 total score.

In sum, through Study 2, we confirmed the factor structure of the HRI-24. Moreover, we provided further support for its good psychometric properties: internal reliability, convergent and divergent validity, and evidence about the HRI-24 total score’s discrimination ability between people at-risk and not at-risk for Hikikomori.

Study 3: Factor and psychometric analyses of the HRI-24 on a Japanese sample

This last study aimed to evaluate if the HRI-24 holds the factor structure and the good psychometric properties found through the previous studies on an East Asian sample and, more specifically, on a Japanese sample of youths and adults.

Methods

Participants

We recruited 392 Japanese (49.7% of females) youths and adults aged between 20 and 50 years (M = 37.11; DS = 7.39). In terms of the highest level of education, 2.0% of the participants completed middle school, 21.4% completed high school, 21.4% had a Bachelor’s Degree, 48.7% had a Master’s Degree, and 5.9% had a Doctoral Degree (0.5% did not answer). Participants were from all regions of Japan, with the majority (30%) from Kanto, 20% from Kansai, 16% from Chubu, 10% from Kyushu, 7% from Tohoku, 7% from Chugoku, 6% from Hokkaido, 3% from Shikoku, and 0.5% from Okinawa (0.5% did not answer). In line with the previous studies, we did not foresee exclusion or inclusion criteria.

Regarding Hikikomori behaviors, the participants reported staying awake after 11 p.m. between 0 and 12 hours (M = 1.57; DS = 1.41) and staying closed in their bedroom between 0 and 24 hours per day (M = 5.43; DS = 5.79).

Measures

Hikikomori Risk Inventory (HRI-24)

In this study, we administered the final version of the HRI-24, which includes questions about Hikikomori typical behaviors. More specifically, we used the Japanese translation, which has been validated through a back-translation process.

NEET-Hikikomori Risk Factors Scale (NHR; Uchida & Norasakkunkit, 2015)

This 27-item self-report scale evaluates tendencies of NEET and Hikikomori. Participants answer on a 7-point Likert scale ranging from “Completely Disagree” to “Completely Agree.” It includes three factors: Freeter Lifestyle Preference, Lack of Self Competence, and Unclear Ambitions for the Future. Since Uchida and Norasakkunkit (2015) did not conduct a CFA on their sample, we conducted a preliminary CFA also on our Japanese participants’ total sample. The 3-factor and 27-item model did not show an optimal fit to the data: χ²/df = 4.82, p < .001; CFI = 0.74; GFI = 0.75; RMSEA = 0.099 (0.094–0.104). Though, except for a reversed item of the Freeter Lifestyle Preference scale (item 7, “I think that a person who does not work will become lazy”), all the items loads on the respective factor at a statistically significant level and with standardized factor loadings ranging between 0.20 (reversed item 5 – Freeter Lifestyle; “I think it is necessary to have a job in order to sufficiently be able to fulfill one’s talents.”) and 0.88. Though, deleting item 7 did not lead to an improvement of the fit: χ²/df = 4.82, p < .001; CFI = 0.75; GFI = 0.76; RMSEA = 0.099 (0.094–0.104). Therefore, we used the original 27-item and 3-factor version of the scale for subsequent analyses, even if we should consider that the NHR does not have a satisfactory fit for this Japanese sample.

Sensation Seeking Scale for Japanese Adolescents (SSS-JA; Shibata, 2008)

This scale was created to measure the interpersonal differences in the optimal level of stimulation. The original English version by Zuckerman (1979) has four factors with 10 items each: Thrill and Adventure Seeking (TAS), Experience Seeking (ES), Disinhibition (Dis), Boredom Susceptibility (BS). When Terasaki et al. (1987) translated the original Sensation Seeking Scale into Japanese, factor analyses showed a different structure with the Japanese sample, where Boredom Susceptibility was not confirmed (Furusawa, 1989; Terasaki et al., 1987). Shibata (2008) performed three tests with university and professional college students aged between 18 and 24, and restructured the scale to fit the cultural context of Japanese youths with the following four factors with five items each: Thrill and Adventure Seeking (TAS; related to new and adventurous activities), Disinhibition (Dis; specifically related to sexual disinhibition), Internal Sensation Seeking (IS; related to the inner worlds of thoughts and imagination), and Daily Novelty Seeking (DNS; such as changing hobbies or doing shopping often). The Cronbach’s alpha for each factor was 0.74 for TAS, 0.69 for Dis, 0.77 for IS, and 0.66 for DNS (Shibata, 2008). We thus used this version with the Japanese sample in our study.

Procedure

Participants were recruited via Lancers, a commonly used website in Japan. Participants were asked to fill out a 10-minute survey online and were rewarded with 150 yen. On the first page of the questionnaire, we told them that we were going to ask some questions regarding their habits, and specified that by filling in the questionnaire on the next pages, the participants gave us their informed consent to take part in the research. Participation was completely voluntary, and participants could withdraw at any time without negative consequences. We guaranteed that data will be kept confidential, and only an ID number would be used to identify each participant in order to receive the payment for their participation. The questionnaire asked participants to provide personal data for demographic questions (e.g., age, gender) before answering the HRI-24, the NHR, and the SSS-JA items. Upon completing the survey, participants were asked to create a code for payment identification and put the same code into Lancers to receive the reward. The study received approval from Nagoya University as part of a series of studies related to Hikikomori research.

Data analysis

We performed analyses using SPSS.25 and AMOS.20. First, we conducted a CFA (Maximum Likelihood estimation) to test if the 24-item and 5-factor solution fits well on the Japanese sample. Hence, we calculated the internal reliability of the five HRI-24 subscales and the total score. Then, we analyzed HRI-24 convergent and divergent validity using Pearson correlations between the HRI-24 and the scales of both the SSS-JA and the NHR. Also, we analyzed the correlation between the HRI-24 and the Hikikomori indicators gathered through the HRI-24 head-sheet on a random subsample of participants, namely about 25% of the total sample (n = 111). Finally, we repeated the previous ANOVAs analyses on the Japanese sample.

Results

First, for the fit of HRI-24 on the total sample (n = 392), we found further support for this model: χ²/df = 2.27, p < .001; GFI = 0.89; CFI = 0.94; TLI = 0.93; RMSEA = 0.057 (0.051–0.063). The standardized factor loadings range between 0.49 and 0.90 (with the exception of item 13, which is associated to a 0.20 factor loading – deleting this item did not improve the goodness of fit), while factor correlations range between 0.27 (Paranoia and Agoraphobia) and 0.64 (Depressive Mood and Anthropophobia). Figure 2 graphically shows this model. All the scales have good internal reliability: Anthropophobia, α = 0.90, item-total correlations between 0.85 and 0.89; Agoraphobia, α = 0.87, item-total correlations between 0.81 and 0.88; Paranoia, α = 0.82, item-total correlations between 0.38 and 0.84; Lethargy, α = 0.82, item-total correlations between 0.72 and 0.88; Depressive Mood, α = 0.91, item-total correlations between 0.79 and 0.90; HRI-24 total score, α = 0.94, item-total correlations between 0.21 and 0.80.

Next, we analyzed the convergent and divergent validity through Pearson correlation between HRI-24 subscales and total score, the SSS-JA, and the NHR subscales (n = 111; see Table 5). We found evidence for convergent validity of the HRI-24, as highlighted by the positive correlations between the HRI-24 total score and subscales and all the NHR scales. Concerning divergent validity, the results showed that both the HRI-24 scales and the HRI-24 total score have negative correlations with the SSS-JA Thrill and Adventure subscale (with some correlations being not statistically significant). The SSS-JA Thrill and Adventure subscale captures the tendency toward extern sensations seeking, which is the opposite of social withdrawal. Concerning the other SSS-JA subscales, instead, there is just a statistically significant correlation between the SSS-JA Disinhibition scale and the HRI-24 Paranoia scale. All the other correlations are not statistically significant.

Table 5.

Convergent and divergent validity analyses of the HRI-24 (Japanese sample, n = 111).

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. HRI-anthropophobia	–
2. HRI-agoraphobia	0.57***	–
3. HRI-paranoia	0.66***	0.35***	–
4. HRI-lethargy	0.54***	0.44**	0.48***	–
5. HRI-depressive mood	0.62***	0.40***	0.61***	0.73***	–
6. HRI-total	0.85***	0.66***	0.82***	0.79***	0.85***	–
7. SSS-thrill and adventure	−0.32***	−0.17	−0.12	−0.26**	−0.23*	−0.27**	–
8. SSS-disinhibition	0.08	0.04	0.20*	0.05	0.04	0.11	0.32***	–
9. SSS-internal sensation S.	−0.04	0.18	0.18	0.15	0.06	0.13	0.18	−0.004	–
10. SSS-daily novelty S.	−0.13	0.02	0.06	0.03	−0.07	−0.02	0.26**	0.33***	0.14	–
11. NHR-Freeter Lifestyle Pref.	0.46***	0.21*	0.43***	0.30**	0.40***	0.46***	−0.08	0.12	−0.03	−0.08	–
12. NHR-Lack Self Comp.	0.77***	0.35***	0.66***	0.56***	0.70***	0.78***	−0.38***	0.03	−0.05	−0.14	0.53***	–
13. NHR-unclear Amb. Future	0.61***	0.27**	0.62***	0.47***	0.71***	0.69***	−0.28**	−0.04	0.02	−0.17	0.45***	0.79***	–
14. Hours/day closed bedroom	0.18	0.09	0.07	0.14	0.11	0.14	−0.15	−0.10	0.07	−0.22*	0.22*	0.29**	0.08	–
15. Hours wake up after 11 p.m.	0.17	0.24**	0.17	0.32***	0.24**	0.28**	0.01	0.17	−0.04	0.08	0.23*	0.24**	0.10	0.34***	–

Note. HRI = Hikikomori Risk Inventory (HRI-24); SSS = Sensation Seeking Scale for Japanese Adolescents; S. = seeking; NHR = Neet Hikikomori Risk Factors Scale; Pref. = preference; Comp. = competence; Amb. = ambition.

*p ⩽ .05. **p ⩽ .01. ***p ⩽ .001.

Finally, concerning the correlations with Hikikomori indicators, the results showed the lack of statistically significant correlations between the HRI-24 and the hours spent closed in the bedroom. Instead, for hours spent awake after 11 p.m. performing activities such as reading or surfing on the Internet, besides the lack of statistically significant correlations with the HRI-24 Antropopohobia and Paranoia subscales, there are positive statistically significant correlations. About these correlations, it should be noted that these two Hikikomori indicators have low correlations with the NHR scales as well.

Next, we repeated the seven ANOVAs performed in Study 2 in order to understand whether the HRI-24 total score distinguishes well people at high risk of Hikikomori (i.e., participants whose total score is above the 75° percentile, that is, between 77 and 115, n = 99) from people with a low risk of Hikikomori (i.e., participants whose total score is below the 25° percentile, namely, between 26 and 53, n = 99). By means of these analyses, we found support for the usefulness of the HRI-24 total score for distinguishing between people at-risk of Hikikomori from others. More specifically, those who have a high HRI-24 score have higher self-reported time spent closed in their bedroom and staying awake after 11 p.m., as compared to people with low HRI-24 scores. Moreover, concerning age, people with a lower risk of Hikikomori are older than people with a high Hikikomori risk. Table 3 shows the results of these analyses. Also, while there is no statistically significant difference between males and females concerning the HRI-24 total score, we found that not having interest in social relationships, preferring virtual relationships to real ones, and having reduced/abandoned activities in the last 6 months is associated with a higher HRI-24 score. Table 4 shows the results of these analyses.

Hence, through Study 2, we provided support for the factor structure of the HRI-24 on a Japanese sample. Moreover, the results showed that the HRI-24 has good psychometric properties on the Eastern sample we analyzed: it has good internal reliability, convergent, and divergent validity. The HRI-24 total score distinguishes well between people at-risk and not at-risk for Hikikomori.

Discussion

The present research aimed to create a self-report instrument for the screening of the general population, and more specifically for detecting people at risk for Hikikomori, namely the Hikikomori Risk Inventory (HRI-24), and to investigate its psychometric properties on both Western (i.e., Italian) and Eastern (i.e., Japanese) populations.

In Study 1, which has been conducted on a sample of Italian adolescents, youths, and adults, we reduced the initial pool of 79 items to the final 24-item and 5-factor version. In Study 2 and Study 3, we deepened the analyses of the psychometric properties of the HRI-24 on both Western (i.e., Italy) and Eastern (i.e., Japan) samples of youths and adults. We found evidence for the good and stable factor structure of the HRI-24 across the two samples (CFI = 0.94, RMSEA = 0.057 for both adult populations). Also, the HRI-24 has good internal reliability for each of its five subscales (i.e., Anthropophobia, Agoraphobia, Paranoia, Lethargy, and Depressive Mood), whose alpha values are all higher than 0.80, and for the total score (α = 0.92 in the Italian sample, α = 0.94 in the Japanese sample).

In addition, using the NEET-Hikikomori Risk Factors Scale (NHR; Uchida & Norasakkunkit, 2015), we found support for the convergent validity of the HRI-24 on both samples. We used a modified NHR version for the Italian participants, which includes 16 items and four scales, as suggested by the Exploratory and Confirmatory Factor Analyses we conducted. Using this version, we found good and positive correlations between the HRI-24 total score and its subscales and the NHR-16 subscales, except for the Freeter Lifestyle Preference scale, which is intended to evaluate the conscious choice of not engaging in full-time employment despite job availability. This NHR-16 scale has a low and positive correlation with the HRI-24 total score, while the correlations with the HRI-24 subscales are generally low and not statistically significant. We speculate that this result might be in line with the fact that the Freeter Lifestyle Preference scale might not be culturally appropriate for Italy’s current economic situation, where there is a lack of full-time permanent jobs. Consequently, in the Italian context, a Freeter Lifestyle Preference would not be perceived as a realistic option, since there is a lack of job availability, which constitutes a premise for being willing to not working full-time despite the possibility to do so. According to recent data, on January 2020 (before the Covid-19 outbreak in Italy), the number of Italian employees decreased by 40,000 units (−0.2% compared to December 2019), and the employment rate was of 59.1%, with these data applying to both women and men, and to both employed and independent contractors (Italian National Institute of Statistics [ISTAT], 2020). In Japan, instead, the employment rate of university graduates was 97.4% in March 2019, and 98% in March 2020 (Japanese Ministry of Health, Labour and Welfare, 2020).

For the Japanese sample, using the original 27-item and 3-factor version of the NHR, we found good and positive correlations between the HRI-24 total score and subscales and the NHR scales (despite the low correlation between the HRI-24 Agoraphobia scale and the NHR Freeter Lifestyle Preference). In this sample, the lowest correlations are also the ones with the Freeter Lifestyle Preference scale, showing that occupational withdrawal may start playing a smaller role in Hikikomori in the current Japanese society as well. The highest correlations are the ones with the Lack of Self Competence scale, which comprehends the items on social issues in the original version.

Concerning divergent validity, again, we found support in both samples. Even if we referred to sensation seeking as the construct that could provide evidence for divergent validity for both the Italian and the Japanese samples, we used two different instruments to evaluate it. Sensation seeking, being characterized by the search for new experiences and sensations (i.e., openness to experiences), might be considered the opposite behavior of Hikikomori, which is characterized by extreme social withdrawal (i.e., avoidance of experiences). For the Italian sample, we administered the Brief Sensation Seeking Scale (BSSS; Hoyle et al., 2002) that foresees a total score only. The results showed that both the HRI-24 total score and its five subscales have no statistically significant correlations with the BSSS. In the Japanese sample, we used the Sensation Seeking Scale for Japanese Adolescents (SSS-JA; Shibata, 2008). The results provided evidence for negative (and generally statistically significant) correlations between the HRI-24 and the SSS-JA Thrill and Adventure scale. Hence, this supported the divergent validity of the HRI-24 since this is the scale that is most related to the opposite behavior of social withdrawal. The other SSS-JA subscales have no statistically significant correlations with the HRI-24, just like in the Italian study (except for the low correlation between the HRI-24 Paranoia scale and the SSS Disinhibition scale; r = 0.20, p = .04). Moreover, it is interesting to note that, even if not statistically significant, some HRI-24 scales and the total score have (low) positive correlations with the Internal Sensation Seeking scale, the scale evaluating some kind of inward withdrawal that could be present in Hikikomori.

Concerning the correlations between the HRI-24 and two Hikikomori behaviors, namely the time spent closed in the bedroom and the time spent awake after 11 p.m., we found few statistically significant (positive) correlations in both the Italian and the Japanese sample. In the Italian sample, these results might be explained by the fact that most participants did not seem to have understood what we meant with these questions, especially concerning the question about the time spent closed in the bedroom. Indeed, in the Italian sample, many participants reported the hours they spent sleeping.

However, it should be noted that both Italian and Japanese participants with high Hikikomori risk (75° percentile on the HRI-24 total score) reported staying longer closed in their bedroom and awake after 11 p.m., as compared to people with low Hikikomori risk (25° percentile on the HRI-24 total score). Moreover, the participants saying not having an interest in social relationships, preferring virtual relationships to real ones, and having reduced/abandoned activities in the last 6 months have a higher HRI-24 total score than people answering in the opposite direction these yes/no questions.

Hence, we speculate that the self-reported hours for the two questions about Hikikomori behaviors might be misleading to detect Hikikomori people. The questions’ misunderstanding might have caused the low and not statistically significant correlations with the HRI-24 scales. Though, the very low and very high end of the answers might be good indicators of the presence/absence of Hikikomori risk, as highlighted by the ANOVAs analyses performed using the 25° and 75° percentile of the HRI-24 total score distribution. Future studies conducted with the modified version of the head-sheet of the HRI-24 could further evaluate their correlations with the HRI-24 scales.

Concerning demographic differences, we found gender differences in neither the Italian nor the Japanese sample, suggesting that this clinical condition might be equally frequent in both genders. However, for age, while there was no statistically significant difference in the Italian sample, there was in the Japanese sample. More specifically, in our Japanese sample, Hikikomori seems to be more common among younger people. Further epidemiological analyses should be conducted on larger samples, especially with younger participants in both Western and Eastern countries.

There are some limits in the present research. First, there were no adolescent samples in both Study 2 and Study 3, although preliminary analyses aimed at establishing the final HRI-24 version included Italian adolescents. Second, for convergent validity of the Italian sample, we used a modified version of the NHR, which evaluates both NEET and Hikikomori tendencies; moreover, the Confirmatory Factor Analysis conducted on the original Japanese sample for the NHR did not highlight a good fit. Finally, since the HRI-24 evaluates the risk for Hikikomori based on the assessment of the presence of psychopathological conditions typically associated with Hikikomori, it might fail to detect the risk for Primary Hikikomori, that is, the ones who have Hikikomori without having other psychological disorders, and who are more spread in Eastern countries, such as Japan (Koyama et al., 2010). Future studies could deepen the psychometric analysis of the HRI-24 on younger samples, as well as in other Western and Eastern countries, and possibly include Hikikomori cases in order to perform Receiver Operating Characteristic (ROC) analyses.

Besides these limits, the present study has provided the scientific literature with a new instrument for evaluating Hikikomori risk, focusing on its psychometric properties in both Western and Eastern samples. Moreover, it is the first Hikikomori instrument to have been properly evaluated in its factor structure through Confirmatory Factor Analyses, which provided support for its factor stability across two Italian samples and one Japanese sample. Hence, future studies could use the HRI-24 to evaluate the Hikikomori risk, and especially for analyzing culture-related features that might characterize Hikikomori across different countries. Finally, the HRI-24 might be used in future studies to analyze the relationship between the social withdrawal of Hikikomori and the social issues that characterize Heavy Study Investment, both in its positive (i.e., Study Engagement) and negative (i.e., Studyholism, or obsession toward study; Loscalzo & Giannini, 2017, 2018a, 2018b) form, as recently shown by Loscalzo and Giannini (2019). In fact, Loscalzo and Giannini (2019) found that both Studyholism and Study Engagement predict family and friends’ complaints and self-reported social impairment due to overstudying. In line with this, they also found that students characterized by high levels of both Studyholism and Study Engagement (i.e., Engaged Studyholics) have higher social impairment than the ones having high Studyholism but low Study Engagement (i.e., Disengaged Studyholics), suggesting that the co-presence of the two types of Heavy Study Investment is associated with higher social impairment. Hence, since Heavy Study Investors are characterized by social withdrawal due to their overstudying behaviors, we speculate that Heavy Study Investment might be a precursor of Hikikomori – which is characterized by social withdrawal as well – especially in countries in which having an outstanding academic career is crucial, both from an individual (Western societies) or collectivistic (Eastern societies) point of view.

In conclusion, this study might provide a new scale to be used jointly with other scales to evaluate emerging psychological disorders across cultures, such as Hikikomori and Studyholism (Loscalzo, 2019).

Footnotes

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Yura Loscalzo

References

Adams

Z. W.

Kaiser

A. J.

Lynam

D. R.

Charnigo

R. J.

Milich

(2012). Drinking motives as mediators of impulsivity-substance use relation: Pathways for negative urgency, lack of premeditation, and sensation seeking. Addictive Behaviors, 37, 848–855.

Aguglia

Signorelli

Pollicino

Arcidiacono

Petralia

(2010). Hikikomori phenomenon: Cultural bound or emergent psychopathology? Giornale Italiano Di Psicopatologia, 16, 1–8.

Aluja

García

L. F.

(2003). Relationships among extraversion, openess to experience, and sensation seeking. Personality and Individual Differences, 35(3), 671–680.

Bentler

P. M.

Bonnet

D. C.

(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588–606.

Chan

G. H.-Y

T. W.

(2014). Hidden youth services: What Hong Kong can learn from Japan. Children and Youth Service Review, 42, 118–126.

De Luca

. (2017). Hikikomori: Cultural idiom or present-day expression of the distress engendered by the transition from adolescence to adulthood. L’Évolution Psychiatrique, 82(1), e1–e15.

De Michele

Caredda

Delle Chiaie

Salviati

Biondi

. (2013). Hikikomori (ひきこもり): A culture-bound syndrome in the web 2.0 era. Rivista di Psichiatria, 48, 354–358.

Funakoshi

Miyamoto

(2015). Significant factors in family difficulties for fathers and mothers who use support services for children with hikikomori. Psychiatry and Clinical Neuroscience, 69(4), 210–219.

Furlong

(2008). The Japanese hikikomori phenomenon: Acute social withdrawal among young people. The Sociological Review, 56(2), 309–325.

10.

Furusawa

(古澤照幸 [1989]). 刺激欲求尺度・抽象表現項目版 (Sensation Seeking Scale-Abstract Expression). 作成の試み. 心理学研究 [An attempt to construct the Sensation Seeking Scale-Abstract Expression]. The Japanese Journal of Psychology, 60(3), 180–184.

11.

García-Campayo

Sanz Abós

Alda

Sobradiel

(2007). Un caso de hikikomori en España [A case report of Hikikomori in Spain]. Medicina Clinica, 129(8), 318–319.

12.

Genda

(2005). A nagging sense of job insecurity: The new reality facing Japanese Youth. International House of Japan, Inc.

13.

Giannini

Loscalzo

(2020). Sensation seeking. In Carducci

B. J.

Nave

C. S.

Saklofske

Stough

Di Fabio

A. M.

(Eds.), The Wiley encyclopedia of personality and individual differences (Vol. III, pp. 1463–1467). John Wiley & Sons.

14.

Guedj-Bourdiau

M.-J.

(2011). Claustration à domicile de l’adolescent. Hikikomori. [Home confinement of the adolescent. Hikikomori]. Annales Médico-Psychologiques, Revue Psychiatrique, 169(10), 668–673.

15.

Hattori

(2006). Social withdrawal in Japanese youth: A case study of thirty-five hikikomori clients. Journal of Trauma Practice, 4, 181–201.

16.

Hoyle

R. H.

Stephenson

M. T.

Palmgreen

Lorch

E. P.

Donohew

R. L.

(2002). Reliability and validity of a brief measure of sensation seeking. Personality and Individual Differences, 32, 401–414.

17.

Istituto Nazionale di Statistica – Italian National Institute of Statistics (ISTAT). (2020). Occupati e disoccupati. Gennaio 2020. Dati provvisori [Employed and unemployed. January 2020. Provisional data]. Retrieved September 28, 2020, from https://www.istat.it/it/files//2020/03/CS_Occupati-e-disoccupati_GENNAIO_2020.pdf

18.

Japanese Cabinet Office. (2019). https://www8.cao.go.jp/youth/kenkyu/life/h30/pdf/s3.pdf

19.

Japanese Ministry of Health, Labour and Welfare. (2020). Retrieved October 22, 2020, from https://www.mhlw.go.jp/stf/newpage_11810.html

20.

Kaneko

(2006). Japan’s ‘socially withdrawn youths’ and time constraints in Japanese society: Management and conceptualization of time in a support group for hikikomori. Time & Society, 15, 233–249.

21.

Kato

T. A.

Tateno

Shinfuku

Fujisawa

Teo

A. R.

Sartorius

Akiyama

Ishida

Choi

T. Y.

Balhara

Matsumoto

Umene-Nakano

Fujimura

Chang

Shadloo

Ahmed

Lerthattasilp

Kanba

(2012). Does the ‘hikikomori’ syndrome of social withdrawal exist outside Japan? A preliminary international investigation. Social Psychiatry and Psychiatric Epidemiology, 47, 1061–1075.

22.

Kato

T. A.

Teo

A. R.

Tateno

Watabe

Kubo

Kanba

(2017). Can Pokémon GO rescue shut-ins (hikikomori) from their isolated world? Psychiatry and Clinical Neuroscience, 71, 75–76.

23.

Kondo

Iwazaki

Kobayashi

Miyazawa

(2007). Seinenki hikikomori keesu no seishin igakuteki haikei ni tsuite [The psychiatric background of social withdrawal in adolescence]. Seishin Shinkeigaku, 109(9), 834–843.

24.

Kondo

Sakai

Kuroda

Kiyota

Kitabata

Kurosawa

(2013). General condition of hikikomori (prolonged social withdrawal) in Japan: Psychiatric diagnosis and outcome in mental health welfare centres. International Journal of Social Psychiatry, 59, 79–86.

25.

Koyama

Miyake

Kawakami

(2010). Lifetime prevalence, psychiatric comorbidity and demographic correlates of ‘hikikomori’ in a community population in Japan. Psychiatry Research, 176, 69–74.

26.

Krieg

Dickie

J. R.

(2013). Attachment and hikikomori: A psychosocial developmental model. International Journal of Social Psychiatry, 59, 61–72.

27.

Lee

Y. S.

Lee

J. Y.

Choi

T. Y.

Choi

J. T.

(2013). Home visitation program for detecting, evaluating and treating socially withdrawn youth in Korea. Psychiatry and Clinical Neuroscience, 67, 193–202.

28.

T. M.

Wong

P. W.

(2015). Youth social withdrawal behavior (hikikomori): A systematic review of qualitative and quantitative studies. Australian and New Zeland Journal of Psychiatry, 49(7), 595–609.

29.

Loscalzo

(2019). Heavy study investment in college students: Studyholism and study engagement prevalence. Applied Psychology Bulletin, 67(286), 55–60.

30.

Loscalzo

Giannini

(2017). Studyholism or study addiction? A comprehensive model for a possible new clinical condition. In Columbus

A. M.

(Ed.), Advances in psychological research (pp. 19–37). Nova Publisher.

31.

Loscalzo

Giannini

(2018a). Problematic overstudying: Studyholism or study addiction? Commentary on: Ten myths about work addiction. Journal of Behavioral Addictions, 7(4), 867–870.

32.

Loscalzo

Giannini

(2018b). Response to: Theoretical and methodological issues in the research on study addiction with relevance to the debate on conceptualising behavioural addictions: Atroszko (2018). Psychiatria i Psychologia Kliniczna, 18(4), 426–430.

33.

Loscalzo

Giannini

(2019). Heavy study investment in Italian college students. An analysis of Loscalzo and Giannini’s (2017) studyholism comprehensive model. Frontiers in Psychiatry, 10, 489.

34.

Maia

Figueiredo

Ponnié-Dax

Vellut

(2014). Hikikomori, ces adolescents en retrait. Armand Colin.

35.

Malagón-Amor

Á.

Córcoles-Martínez

Martín-López

L. M.

Pérez-Solà

. (2015). Hikikomori in Spain: A descriptive study. International Journal of Social Psychiatry, 61, 475–483.

36.

Ovejero

Caro-Cañizares

de León-Martínez

Baca-Garcia

(2014). Prolonged social withdrawal disorder: A hikikomori case in Spain. International Journal of Social Psychiatry, 60(6), 562–565.

37.

Pokhrel

Sussman

Sun

Kniazer

Masagutov

(2010). Social self-control, sensation seeking and substance use in samples of US and Russian adolescent. American Journal of Health Behavior, 34(3), 374–384.

38.

Primi

Narducci

Benedetti

Donati

Chiesi

(2011). Validity and reliability of the Italian version of the Brief Sensation Seeking Scale (BSSS) and its invariance across age and gender. Testing, Psychometrics, Methodology in Applied Psychology, 18(4), 231–241.

39.

Ranieri

(2015). When social withdrawal in adolescence becomes extreme: The “hikikomori” phenomenon in Italy. Psychiatria i Psychologia Kliniczna, 15(3), 148–151.

40.

Ranieri

(2016). Adolescenti tra abbandono scolastico e ritiro sociale: Il fenomeno degli <<hikikomori>> [Adolescents between school dropout and social withdrawn: The <<hikikomori>> phenomenon]. Psicologia Clinica dello Sviluppo, 2, 319–326.

41.

Ranieri

Andreoli

Bellagamba

Franchi

Stoppielli

(2015). Adolescenti tra abbandono scolastico e ritiro sociale: Il fenomeno degli “hikikomori” ad Arezzo. [Adolescents between school dropout and social withdrawal: The phenomenon of “hikikomori” in Arezzo]. Il Cesalpino, 14, 13–17.

42.

Reeve

B. B.

Hays

R. D.

Bjorner

J. K.

Cook

K. F.

Crane

P. K.

Teresi

J. A.

Thissen

Revicki

Weiss

Hambleton

Liu

Gershon

Reise

Lai

J.-S.

Cella

(2007). Psychometric evaluation and calibration of health related quality of life item banks: Plans for the patient-reported outcomes measurement information system (PROMIS). Medical Care, 45(5), S22–S31.

43.

Saitō

(1998). Shakaiteki Hikikomori: Owaranai Shishunki [Social withdrawal: A never ending adolescence]. PHP Kenkyujo.

44.

Saitō

(2010). Hikikomori no hyouka shien ni kansuru gaidorain [Guideline on evaluation and support of hikikomori]. Ministry of Health, Labour & Welfare.

45.

Sakamoto

Martin

R. G.

Kumano

Kuboki

Al-Adawi

(2005). Hikikomori, is it a culture-reactive or culture-bound syndrome? Nidotherapy and a clinical vignette from Oman. International Journal of Social Psychiatry, 35(2), 191–198.

46.

Sarchione

Santacroce

Corbo

Lupi

Marinotti

Di Giannantonio

(2014). Hikikomori, clinical and psychopathological issues. European Psychiatry, 30, 28–31.

47.

Schermelleh-Engel

Moosbrugger

Muller

(2003). Evaluating the fit of structural equation models: Tests of significance and goodness-of-fit models. Methods of Psychological Research – Online, 8, 23–74.

48.

Shibata

(柴田由己 [2008]). 青年用刺激希求尺度の信頼性・妥当性の検 [Examination of the reliability and validity of the Sensation Seeking Scale for Adolescents]. The Japanese Journal of Personality, 16(2), 198–208.

49.

Smith

(2015). Stagnant youth: The NEET, freeter, and hikikomori phenomena. Post Bubble Culture: Research on Contemporary Japan. Retrieved 12 March 2020, from http://postbubbleculture.blogs.wm.edu/category/postings_gnc/page/2/

50.

Stautz

Cooper

(2013). Impulsivity-related personality traits and adolescent alcohol use: A meta-analytic review. Clinical Psychology Review, 33(4), 574–592.

51.

Stip

Thibault

Beauchamp-Chatel

Kisely

(2016). Internet addiction, hikikomori syndrome, and the prodromal phase of psychosis. Frontiers in Psychiatry, 7, 6.

52.

Suwa

Suzuki

(2013). The phenomenon of “hikikomori”(social withdrawal) and the socio-cultural situation in Japan today. Journal of Psychopathology, 19, 191–198.

53.

Suwa

Suzuki

Hara

(2003). Family features in primary social withdrawal among young adults. Psychiatry and Clinical Neuroscience, 57, 586–594.

54.

Tajan

(2015). Social withdrawal and psychiatry: A comprehensive review of hikikomori. Neuropsychiatrie de l’Enfance et de l’Adolescence, 63(5), 324–331.

55.

Tajan

Yukiko

Pionniè-Dax

(2017). Hikikomori: The Japanese Cabinet Office’s 2016 Survey of Acute Social Withdrawal. The Asia-Pacific Journal, 15(5), 5017.

56.

Tateno

Park

T. W.

Kato

T. A.

Umene-Nakano

Saitō

(2012). Hikikomori as a possible clinical term in psychiatry: A questionnaire survey. BMC Psychiatry, 12, 169.

57.

Taylor

(2006). Critical policy evaluation and the questions of values: A psychosocial approach. Critical Social Policy, 26(1), 243–267.

58.

Teo

A. R.

(2010). A new form of social withdrawal: A review of hikikomori. International Journal of Social Psychiatry, 56, 178–185.

59.

Teo

A. R.

(2012). Social isolation associated with depression: A case report of hikikomori. International Journal of Social Psychiatry, 198, 339–341.

60.

Teo

A. R.

Chen

J. I.

Kubo

Katsuki

Sato-Kasai

Shimokawa

Hayakawa

Umene-Nakano

Aikens

Kanba

Kato

T. A.

(2018). Development and validation of the 25-item Hikikomori Questionnaire (HQ 25). Psychiatry and Clinical Neurosciences, 72, 780–788.

61.

Teo

A. R.

Fetters

M. D.

Stufflebam

Tateno

Balhara

Choi

T. Y.

Kanba

Mathews

C. A.

Kato

T. A.

(2015). Identification of the hikikomori syndrome of social withdrawal: Psychosocial features and treatment preferences in four countries. International Journal of Social Psychiatry, 61, 64–72.

62.

Teo

A. R.

Gaw

A. C.

(2010). Hikikomori, a Japanese culture-bound syndrome of social withdrawal? The Journal of Nervous and Mental Diseases, 198, 444–449.

63.

Terasaki

Shiomi

Kishimoto

Hiraoka

(寺崎正治・塩見邦雄・岸本一・岡清志 [1987]). 日本語版 [Sensation-Seeking Scale’. の作成 [A Japanese version of the Sensation-Seeking Scale]. The Japanese Journal of Psychology, 58(1), 4–48.

64.

Toivonen

Norasakkunkit

Uchida

(2011). Unable to conform, unwilling to rebel? Youth, culture, and motivation in globalizing Japan. Frontiers in Psychology, 2, 207.

65.

Uchida

(2010). Apathetic and withdrawing students in Japanese universities―with regard to hikikomori and student apathy. Journal of Medical and Dental Sciences, 57(1), 95–108.

66.

Uchida

Norasakkunkit

(2015). The NEET and Hikikomori spectrum: Assessing the risks and consequences of becoming culturally marginalized. Frontiers in Psychology, 6, 1117.

67.

Umeda

Kawakami

(2012). World Mental Health Japan Survey Group 2002–2006. Association of childhood family environments with the risk of social withdrawal (‘hikikomori’) in the community population in Japan. Psychiatry and Clinical Neuroscience, 66, 121–129.

68.

Watts

(2002). Public health experts concerned about ‘hikikomori’. The Lancet, 359, 1131.

69.

Zuckerman

(1979). Sensation seeking: Beyond the optimal level of arousal. Lawrence Erlbaum Associates.

70.

Zuckerman

Eysenck

S. B. G.

Eysenck

H. J.

(1978). Sensation seeking in England and America: Cross-cultural, age, and sex comparisons. Journal of Consulting and Clinical Psychology, 46, 139–149.