Work addiction

Based on three representative population studies conducted in Norway, Hungary, and Germany, 8–10% of the employed population is affected by work addiction in the western culture [1–3]. Moreover, the prevalence of work addiction in South Korea is much higher, namely 39.7% [4]. In alignment with this high populational involvement, the topic of work addiction is in the limelight of research [5]. Initially, Oates [6] was using the term ‘workaholism’ in the early 1970s. He proposed that workaholism is driven by an uncontrollable need to work constantly, so the affected individual is continuously thinking about work, even when working hours are over, and the person works more than the employer usually expects [6]. Since then, the term ‘workaholic’ has become a frequently used phrase describing individuals who work ‘too much’. On the other hand, researchers generally use both terms [workaholism and work addiction] interchangeably, and as such it shows the confusion and lack of consensus of definition. Griffiths and colleagues [5] suggested differentiating between workaholism and work addiction, namely, ‘workaholism’ is a term that appears in the literature in a more positive way [5], and it was also defined as the „best-dressed mental health problem of the twentieth century” [7, p. 2]. Hence “work addiction” is a more concrete psychological construct and it fits more into the addiction framework [5]. Research on behavioral addictions has made significant progress in recent decades, contributing to the conceptual validity of certain disorders such as gambling disorder and gaming disorder [8]. Work addiction is often referred to as a behavioral addiction, however, it is still not included in the DSM-5 [9], or the ICD-11 [10].

According to the existing literature, research on work addiction has been examined from different approaches, such as by measuring time spent with work [11] or inner drives and needs [12]. Ng and colleagues [13] defined work addiction as a behavioral pattern that includes excessive amount of time and effort investment in work in combination with obsession with work. According to them, individuals who are at risk of work addiction tend to devote extensive hours to work, often sacrificing their personal time as a result of their compelling obsession. A new conceptualization of work addiction that encompasses cognitive, emotional, motivational, and behavioral dimensions of the problem was proposed by Clark and colleagues [14]. These authors described work addiction with the following four elements 1) compulsive motivation to work (i.e., motivational factor), 2) thinking of work persistently and in an uncontrollable way (i.e., cognitive factor), 3) the presence of negative emotions if working is not possible (i.e., emotional factor), and 4) excessive work that is more than expected or required (i.e., behavioral factor) [14]. The strength of this new concept is emphasizing the main psychological processes in the background of work addiction, however, this dimensional model does not include all the symptoms regarding the addictive nature of the problem. According to Griffiths and Karanika-Murray [15], work addiction shows the same components as other substance-related and other addictions, namely: 1) salience, 2) tolerance, 3) mood modification, 4) relapse, 5) withdrawal, 6) intrapersonal and interpersonal conflicts, and 7) problems caused by the compulsive behavior.

Work addiction and social relationships in theoretical models

The negative impact of work addiction on one’s social relationships has been highlighted for a long time. It was already stated in Oates’ early definition that workaholics are those people for whom “need for work has become so excessive that it creates noticeable disturbance or interference with his bodily health, personal happiness, and interpersonal relations, and with his smooth social functioning” [6, p.4.]. This definition is in line with the components model of addictions [16, 17] that emphasizes impaired social functioning regarding two components: 1) "intrapersonal conflicts” refer to difficulties in balancing the individual’s professional and personal life ending in regular conflicts with family members, relatives, friends, and important others; and 2) “problems” reflects the neglect of social life or losses of important relationships because of excessive and compulsive work. These theoretical concepts are in accordance with the clinical therapist Robinson’s observations who was one of the first authors to develop a measure for assessing the risk of work addiction [18]. Based on his family therapist experiences, Robinson [19] listed the “difficulty with relationships” as an indicator of work addiction in his conceptualization. As Robinson [7] argued, as thoughts and energies of a person suffering from work addiction are always related to work, people surrounding them tend to feel neglected due to the lack of attention they receive. As a result, individuals with work addiction have poorer family relationships, and family members of workaholics more often show symptoms of mental problems [20]. Years later, Schaufeli and colleagues [21] developed a model of work addiction that highlighted two primary components: excessive work and compulsive work. According to the model, both excessive and compulsive work exert a significant and negative influence on personal life and social relationships. Workaholics tend to spend significantly more time at work than with friends, engaging in entertainment, or attending social events. They find it particularly challenging to allocate their free time away from work.

In another comprehensive model of work addiction, Loscalzo and Giannini [22] proposed that work addiction is a clinical condition containing internalizing and externalizing symptoms, and these authors suggested using DSM-like criteria for work addiction, as with other addictive disorders. Among the twelve criteria, three contain problems in social functioning, namely: 1) „recurrent work-related behaviors as described above resulting in a failure to fulfill major role obligations at home”; 2) „continued work-related behaviors as described above despite having persistent or recurrent social or interpersonal problems caused or exacerbated by these behaviors themselves”; and 3) „important social, family, or recreational activities are given up, reduced or impaired because of workaholism” [22, p. 321.]. Taking a closer look at the empirical models that emphasize strong connection between work addiction and difficulties in social functioning, we should also mention the transactional model [23]. This model argues that coping as one’s choice is determined by primary appraisals and secondary appraisals. McMillan and colleagues [24] formulated the coping model of work addiction, and they highlighted that the partner of the individual with work addiction could act as a stress-buffer, helping the person to cope with stress factors. In addition, they suggested that work addiction can be characterized by denial and relationship distress as well [24], thus work addiction has damaging side effects on one’s health and social relationships. Research on family systems aims to develop models on the relationship between family functioning and dysfunctional behavior (i.e., work addiction), and these models suggest that the problem is located rather in the family system than in the person [24].

During the last decades, several research investigated the correlates, risk factors, and possible consequences of work addiction. At the same time, most of the theoretical concepts of work addiction emphasize the harmful effect of work addiction on the quality of social life. Although one meta-analytic study has already investigated the relevant correlates of work addiction including some of the social factors [25], a comprehensive review that incorporates all the social life characteristics of work addiction is still missing. The meta-analysis conducted by Clark et al. [25] focused only on a narrower aspect of the research question. Their aim was to comprehensively analyze the correlates of work addiction, which led them to include only certain aspects of social relationships (i.e., family satisfaction/functioning, marital disaffection, and work-life conflict). It is important to note that their search concluded eleven years ago, in 2013, and thus many recent articles have not been included in their analysis. Additionally, the authors did not provide specific details about each individual article included, making it unclear how many relevant studies on social relationships were used for their analysis. However, it is known that the meta-analysis included three articles to analyze marital disaffection, six articles to analyze relationship satisfaction, and 22 articles to analyze work-life conflict. Our review, on the other hand, aimed to incorporate all empirical work relevant to social relationships and analyze it in a narrative manner. By adopting this approach, we can obtain a more up-to-date and comprehensive understanding of the social relationship experiences of individuals affected by work addiction.

Aim of the current study

One of the aims of the present paper was to systematically review the literature that examined any dimensions of social life in connection with work addiction. This is planned to be explored both in the broader sense, which includes issues of work-life balance and social functioning, but also at a more concrete level, which refers to specific social relations. In the latter, we plan to examine all social relationships, including family life and couple relationships in the private sphere, as well as broader community, friendship, and workplace relationships. Thus, our aim is to examine six broad themes: (1) work-life balance, (2) general social functioning, (3) actual family life, (4) family of origin and offspring of workaholics, (5) intimate relationship, and (6) other relationships i.e., relationships with friends, community, and colleagues. Our further aim was to conduct meta-analyses of the results published so far. Based on the theoretical models of work addiction [7, 13, 14, 17, 21, 22], we expected that individuals affected by work addiction have more conflicts between work and life, work and family, and have poorer social relationships. In sum, the main goal of the current systematic review is to compare research findings with empirical models, and to verify or reject the assumption on the relationship between work addiction and problematic social life.

Operational definitions

Studies concerning a wide range of focal points were included, since we aimed to examine the relationship between work addiction and different constructs regarding social relationships and social life. To analyze the vast field of social relationships in a more manageable manner, we have categorized them into six themes. This approach involves a progressively narrower focus, starting with the broader category of private life (i.e., balance between work and life), followed by general social functioning, then specific social relationships. Thus, we created six main categories for the social relationship variables, namely (1) work-life balance, (2) general social functioning, (3) actual family life, (4) family of origin and offspring of workaholics, (5) intimate relationship, and (6) other relationships i.e., relationships with friends, community, and colleagues. In S2 Appendix, we summarize the concepts that were included in the six themes.

Search strategy

A computerized literature search was performed to collect all the relevant papers published (at any time from inception until December 2022). We conducted a systematic literature search using five databases: Web of Science, Science Direct, PsycInfo, EBSCO, and PubMed. These databases are extensively used by authors who publish systematic review and meta-analysis studies. From the perspective of work addiction research, these databases are considered the most relevant ones. While searching in the databases, we used the following keywords: “work addiction” OR “workaholism” OR “workaholic” AND “social” OR “social relationship” OR “friend” OR “family” OR “familial” OR “marriage” OR “marital” OR “married” OR “spouse” OR “spousal” OR “divorce” OR “work-life” OR “child” OR “parent” OR “parental” OR “mother” OR “father” OR “wife” OR “husband” OR “colleague” OR “supervisor”. In addition, the reference lists of all articles detected via our search were scanned for further references.

Eligibility criteria and study selection

The study titles and the abstracts were screened for relevance. In case the abstract included information to determine eligibility, we reviewed the full text of the article. To determine which articles should be included for analysis, two of the authors independently evaluated the full-text articles. Inclusion and exclusion process is presented in detail in Fig 1. We included those quantitative studies that provided empirical data on the relationship between work addiction and any variable that assesses social relationships, or studies that investigate the relationship between two generations (i.e., parents and offspring) in the context of work addiction. Studies were included if they met the following criteria: (i) a psychometrically validated work addiction scale was used; (ii) at least one psychometrically validated scale or subscale was used to measure a social variable; (iii) parent-child relationship was investigated by a psychometrically validated scale or subscale from the perspective of work addiction; (iv) the paper was written in English; and (v) any population was involved. The articles we have included may contain population data primarily focused on adults, as work addiction is most relevant to this demographic. However, selected articles also encompass studies involving children, particularly those examining work addiction within the context of parent-child relationships. Studies were excluded, if they were: (i) theses or doctoral dissertations; (ii) non-empirical studies; (iii) non-academic publications (grey literature); (iv) reviews (e.g., theoretical studies, review articles or books reviews); (v) studies, which have not examined any relationship between work addiction and social life; and (vi) studies, which were not written in English. Two authors examined the titles and the abstracts of the identified articles, and disagreements were discussed separately, aiming for a consensus. For details of the inclusion and exclusion process, please review the PRISMA Flow Diagram Fig 1.

In addition to the systematic review, we also planned to conduct meta-analyses, which meant further selection steps. Two additional inclusion criteria were used for inclusion in the meta-analysis: (i) Correlation analysis between work addiction and the social relationship variable; (ii) Analysis of the relationship between the two variables within a single individual. The latter means that we excluded from the analysis those where the association between a person’s work addiction (e.g., supervisor, spouse) and another person’s (e.g., subordinate, partner) social characteristic (e.g., work-family conflict) was analyzed. As the topics of social relationships are extremely diverse, we decided to carry out different meta-analyses according to the previously presented six main themes. Some articles could be included in more than one meta-analysis if they covered more than one topic.

Data extraction

Each study was evaluated according to a specific data extraction structure, which included the following data: (i) bibliographic information (author(s), year of publication and the country where the data were collected); (ii) research design (i.e., cross-sectional, longitudinal); (iii) sample characteristics (mean age, gender ratio, sample size); (iv) work addiction questionnaire (e.g., Work Addiction Risk Test [WART], Dutch Work Addiction Scale [DUWAS], etc.); (v) social relationship variables (e.g., work-life balance, marital satisfaction, etc.); (vi) measurement tool of the social relationship variable (e.g., Work-Family Conflict Scale, Kansas Marital Satisfaction Scale, etc.); (vii) Cronbach’s alpha values for all the questionnaires used; (viii) statistical methods applied (e.g.: correlation analyses, regression analyses, ANOVA, etc.); and (ix) the main results of the statistical analyses. Regarding the results, the correlation coefficients, standardized beta values, and means are given in the summary table. Where specific information was missing from the article, it was marked as not available (n/a). In the case of missing data, we attempted to contact the authors of the primary article. If our contact attempts were unsuccessful, we indicated in the table that no data were available.

Data extraction was executed separately by two authors, and we used the form shown in S3 Appendix. The form was initially tested and agreed upon in a trial run to ensure that everyone had a consistent understanding of each category. We found acceptable agreement between the authors on data extraction. Interrater reliability was high, ranging from 92.54% to 100%. Inconsistencies and disagreements were discussed and resolved via consensus. For assessing the methodological quality of the papers, we applied the Joanna Briggs Institute’s (JBI) critical appraisal tool for cross-sectional studies [27]. This appraisal is a widely recognized method for evaluating the methodology of systematic reviews and meta-analyses, which is why we chose to utilize it. This assessment was also conducted independently by two of the authors, mirroring the article review and data extraction process. Any discrepancies were resolved through separate discussions aimed at reaching a consensus. The JBI critical appraisal tool contains eight items that could be answered with “yes”, “no”, “not applicable (n/a)”, or “unclear”. We calculated a total score for each study by using one point for any affirmative answer, and zero point for any other answers. We also calculated the percentages of yes answers based on the number of available relevant criteria. The results of the risk of bias assessment can be found in S2 Table.

Data analysis

Not all articles were included in the meta-analytical process due to not meeting all the criteria. If any of the six major themes failed to meet the meta-analysis criterion, we conducted a narrative synthesis.

The Comprehensive Meta-Analysis software was used to conduct the analyses. The dependent variable in all the meta-analyses was the correlation coefficient (r) between work addiction and a social relationship variable. In the first three meta-analyses (work-life balance, general social functioning, and family life), higher correlation coefficients indicate an association between work addiction and difficulties in social relationships. However, in the fourth and fifth meta-analyses (intimate relationships and relationships with community, friends, and colleagues), the correlation coefficient represents the association between work addiction and the quality of social relationships.

We evaluated the heterogeneity of the average effect sizes using Q-statistics and estimated I². To assess publication bias, we examined the funnel plot, computed Egger’s test, and employed the Duval and Tweedie trim and fill procedure. Additionally, several moderator analyses were performed in each of the five meta-analyses. Given that many studies reported multiple effect sizes, we combined all effect sizes within each category for categorical moderators. For numerical predictors like gender ratio, mean age of the sample, and study quality, we conducted meta-regression analyses.

Studies included in the systematic review

Fig 1 displays the PRISMA flow diagram illustrating the numbers of excluded and included studies. In total, 1,659 articles were identified and reviewed based on eligibility and exclusion criteria. As a result of checking the reference lists of all articles, no new studies (n = 0) were added to the existing list. After removing the duplicates (n = 893) with EndNote (Clarivate Analytics), we continued reviewing the abstracts of the included studies (n = 766). The abstract and language screening resulted in the exclusion of non-relevant records (n = 45). As a next step we assessed the remaining articles (n = 721) with full-text review. During the full-text review, we excluded studies that were doctoral dissertations, non-empirical studies (n = 619), or reviews and studies that did not examine any relationship between work addiction and social life (n = 530) or were not written in English. After completion of the entire screening process, we finalized the list of articles (n = 102) to be included into our systematic review (S1 Table).

Description of the included studies

The 102 included studies were published between 1995 and 2022 (until December 31). Nine studies (8.8%) were published between 1995 and 2000, 19 studies (18.6%) between 2001 and 2010, and almost three quarters of the papers (N = 74; 72,5%) were published between 2011 and 2022. More than one-third of the studies (n = 38) were conducted in Europe and one-third were carried out in North America (n = 35). The remaining studies were conducted in Asia (n = 14), Australia (n = 5), Middle East countries (n = 7), and South America (n = 3). A convenience sample was used in all studies. A total of 15 articles applied longitudinal research design, and all the other studies (n = 87) used cross-sectional design. Most of the studies (n = 99) involved adult participants, and only three studies investigated both parents and their children. The mean age of the adult samples was between 22.14 and 54.7 years. The largest sample included 8,419 participants, and the smallest sample involved 24 individuals. Altogether, the total number of adult participants included in the current systematic review was 66,186, and a total of 333 families were included. The average ratio of males in the studies was 40.37%.

As mentioned above, the included quantitative studies used at least one psychometrically validated work addiction scale. S1 Fig presents all thirteen different work addiction measures applied in the studies and the number of the studies using these instruments.

The quality analysis of the primary studies (S2 Table) showed that the average quality rating of 102 included studies was 88%. About half of the studies (n = 56; 55%) reached the maximum scores (100%) in the checklist, and more than one-fifth of the studies (n = 20; 19.6%) were rated 86%. A total of 14 studies (13.7%) were rated 71.4% on the checklist, and nine of them (8.8%) reached 57.1%. Finally, only three studies (2.9%) did not reach the 50% level of the quality rating, these articles scored 43%.

Meta-analyses

A total of 22 articles were excluded from the meta-analyses due to a lack of correlation analysis, and a further 5 articles were excluded due to the analysis was conducted between a person’s work addiction and another person’s social characteristics. The excluded studies comprised 6 articles on general social functioning, 9 articles on actual family life, 7 on family of origin and offspring of workaholics, 4 on intimate relationships, and 2 on relationships with friends, community, and colleagues.

A total of 75 articles were included in the meta-analytic part of the study. Although six themes were initially identified, and theoretically, six meta-analyses could have been conducted, one of the themes lacked sufficient eligible studies. Consequently, we were able to conduct five meta-analyses. For each meta-analysis, we have included the following number of articles: (1) Work-life balance: 10 studies; (2) General social functioning: 9 studies; (3) Current family life: 47 studies; (4) Intimate relationships: 12 studies; and (5) Community, friendship, and colleagues: 14 studies. Regarding the topic of ‘family of origin and offspring of workaholics’, there were very few studies available, none of which met the inclusion criteria for the meta-analysis. Therefore, we processed the studies on this topic in the form of a narrative synthesis, and the results are presented subsequent to the five meta-analyses.

1. Meta-analysis: Work addiction and work-life imbalance.

Results of 11 studies, including data from 3,415 participants, were synthesized. There were no outliers. The average correlation between work addiction and work-life imbalance was positive, moderate, and significant (r = .338, 95% CI: 0.231; 0.437, p < .001). This was a heterogeneous effect (see S3 Table). The forest plot for the second meta-analysis is shown in Fig 2. Regarding publication bias, Egger’s test was not significant (intercept: 5.48, p = .148), and the funnel plot showed asymmetry (see S2 Fig), however, Duval and Tweedie’s Trim and Fill calculation showed a smaller but significant average effect size (r = .213, 95% CI: 0.085; 0.335). According to Rosenthal’s classic fail-safe n of 995, it was a robust effect.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Forest plot of the first meta-analysis: The relationship between work addiction and work-life imbalance.

https://doi.org/10.1371/journal.pone.0303563.g002

We could analyze the effects of two social relationship variables separately. While work-life enrichment (based on only 1 study) does not show a significant relationship with work addiction (r = –.045, CI: –0.125; 0.035, k = 1, Z = –1.11, p = 0.268), work-life imbalance exhibits a medium and significant positive relationship with work addiction (r = .362, 95% CI: 0.251; 0.463, k = 11, Z = 6.08, p < .001).

2. Meta-analysis: Work addiction and difficulties in general social functioning.

Results of 9 studies, including data from 2,088 participants, were synthesized. There were no outliers. The average correlation between work addiction and difficulties in general social functioning was small but positive and significant (r = .274, 95% CI: 0.166; 0.376, p < .001). This was a heterogeneous effect (see S3 Table). The forest plot for the third meta-analysis is shown in Fig 3. Regarding publication bias, the funnel plot was symmetrical (see S3 Fig) but the Egger’s test was not significant (intercept: 2.79, p = .249). However, Duval and Tweedie’s Trim and Fill calculation showed a significant average effect size (r = .274, 95% CI: 0.165; 0.376). According to Rosenthal’s classic fail-safe n of 285, it was a robust effect.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Forest plot of the second meta-analysis: The relationship between work addiction and difficulties in general social functioning.

https://doi.org/10.1371/journal.pone.0303563.g003

When we analyzed each of the sub-themes within the social functions, we found that work addiction exhibited a significant moderate negative relationship with the quality of social life and a significant positive but weak relationship with social dysfunction and the amount of social conflict (Table 1). However, no significant relationship with the amount of time spent with others and the amount of social support was found for work addiction (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Results of the moderator analyses focusing on the social variables in context of general social life.

https://doi.org/10.1371/journal.pone.0303563.t001

3. Meta-analysis: Work addiction and difficulties in family relationships.

After excluding an outlier [28], we synthesized 49 studies reporting on 37,917 participants in the first meta-analysis. The average correlation between work addiction and difficulties in family-related social relationships was small but positive and significant (r = .284, 95% CI: 0.216; 0.350, p < 001). This effect was heterogeneous (see S3 Table). The forest plot for the first meta-analysis is shown in Fig 4. Regarding publication bias, Egger’s test was significant (intercept: –5.78, p < .001), the funnel plot was symmetrical (see S4 Fig), and Duval and Tweedie’s Trim and Fill calculation showed a significant average effect size (r = .284, 95% CI: 0.216; 0.349). Finally, according to Rosenthal’s classic fail-safe n of 6,719, the effect appeared robust.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Forest plot of the third meta-analysis: The relationship between work addiction and difficulties in family life.

https://doi.org/10.1371/journal.pone.0303563.g004

Analyses were also carried out separately for each of the main social relationship variables. The largest number of studies was on work-family conflict (k = 42), showing a significant positive relationship of moderate strength (Table 2). There is also a significant positive but weak relationship between work addiction and family dysfunction, but only one study supports this finding. A significant, yet weak, negative relationship is found with work-family facilitation, satisfaction with family relationships, and family engagement. Although based on only one study, a significant, moderate-strength, negative relationship is observed between work addiction and family support. However, work addiction shows no relationship with family conflict or the quality of family relationships (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Results of the moderator analyses focusing on the social variables in context of family life.

https://doi.org/10.1371/journal.pone.0303563.t002

4. Meta-analysis: Work addiction and the quality of intimate relationship.

Results of 13 studies, including data from 2,990 participants, were synthesized. There were no outliers. The average correlation between work addiction and the quality of intimate relationship was small but negative and significant (r = –.243, 95% CI: –0.356; –0.123, p < .001). This was a heterogeneous effect (see S3 Table). The forest plot for the third meta-analysis is shown in Fig 5. Regarding publication bias, Egger’s test was not significant (intercept: 2.94, p = .945) and the funnel plot showed asymmetry (see S5 Fig), however, Duval and Tweedie’s Trim and Fill calculation showed a significant average effect size (r = –.315, 95% CI: –0.427; –0.193). According to Rosenthal’s classic fail-safe n of 565, it was a robust effect.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Forest plot of the fourth meta-analysis: The relationship between work addiction and the quality of intimate relationships.

https://doi.org/10.1371/journal.pone.0303563.g005

When analyzed separately for each theme, there is a significant negative relationship of moderate strength between work addiction and relationship quality and positive feelings in the relationship (Table 3). There is also a moderate but positive relationship between work addiction and relationship disaffection. However, there is no significant correlation between relationship satisfaction, relationship support, and work addiction (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Results of the moderator analyses focusing on the social variables in context of the quality of intimate relationships.

https://doi.org/10.1371/journal.pone.0303563.t003

5. Meta-analysis: Work addiction and the quality of relationships with friends, community, and colleagues.

After excluding an outlier [29], we synthesized 14 studies reporting on 9,733 participants. The average correlation between work addiction and the quality of relationships with friends, community, and colleagues was small but negative and significant (r = –.156, 95% CI: –0.218; –0.092, p < .001). This was a heterogeneous effect (see S3 Table). The forest plot for the first meta-analysis is shown in Fig 6. Regarding publication bias, the funnel plot was symmetrical (see S6 Fig) but Egger’s test was not significant (intercept: –2.10, p = .275). However, Duval and Tweedie’s Trim and Fill calculation showed a significant average effect size (r = –.155, 95% CI: –0.218; –0.091). According to Rosenthal’s classic fail-safe n of 584, it was a robust effect.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Forest plot of the fifth meta-analysis: The relationship between work addiction and quality of relationships with friends, community, and colleagues.

https://doi.org/10.1371/journal.pone.0303563.g006

By analyzing each of the social variables separately, we find that while relationship quality and relationship cohesion show a weak but significant negative relationship with work addiction, social support, relationship satisfaction, and conflict levels are not related to work addiction (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Results of the moderator analyses focusing on the social variables in context of the quality of relationships with friends, community, and colleagues.

https://doi.org/10.1371/journal.pone.0303563.t004

Since this meta-analysis included several types of non-family-related social relationship factors, moderator analysis was also conducted for each type of relationship. Again, all relationships remained significantly negative, and although all were weakly correlated, with varying strengths. Thus, the strongest correlation was found between work addiction and community. Additionally, within the workplace, the negative correlation with coworkers was stronger than with the manager (Table 5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Results of the moderator analyses focusing on the social relationship categories in context of the quality of relationships with friends, community, and colleagues.

https://doi.org/10.1371/journal.pone.0303563.t005

Furthermore, a moderator analysis was conducted for all five meta-analyses on the work addiction scales. The results of this analysis are summarized in S4 Appendix.

Meta-regression analyses.

The mean age of the sample served as a significant moderator between work addiction and work-life imbalance, however, the coefficient was very small (Coefficient = –.023; Z = –2.76; p = .005). This result indicates that among younger individuals, the relationship between work addiction and work-life imbalance is higher. Mean age did not serve as a significant moderator in the other four meta-analyses (see Table 6). Concerning the gender ratio of the samples, we observed a significant moderator effect only for work-life imbalance. We found a very small negative effect (Coefficient = –.01; Z = –8.10; p < .001), suggesting that the positive relationship between work addiction and work-life imbalance is stronger among females than males. In the other four meta-analyses, we did not observe any significant moderator effect of the gender ratio (Table 6). Based on the quality scores of the studies, significant effects were found in only one of the five meta-analyses. There exists a positive effect regarding difficulties in family relationships; specifically, higher-quality studies indicate a stronger association between work addiction and challenges in family relationships. It is essential to note, however, that the effect size in this context is very modest (Coefficient = .006; Z = 2.36; p = .018) (Table 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Results of the meta-regression analyses.

https://doi.org/10.1371/journal.pone.0303563.t006