Properties of the Problematic Pornography Consumption Scale (PPCS-18) in community and subclinical samples in China and Hungary (2020)

1. Introduction

Data suggest that increased internet use has been accompanied by increases in pornography consumption and frequencies of problematic pornography use (PPU), representing clinically relevant phenomena (Brand, Antons, Wegmann, & Potenza, 2019a; Brand, Blycker, & Potenza, 2019b; de Alarcón, de la Iglesia, Casado, & Montejo, 2019). Despite an increase in studies on internet-related problems and disorders, conceptualizations of PPU remain debated (Hertlein and Cravens, 2014, López-Fernández, 2015, Potenza et al., 2017, Stark et al., 2018, Wéry and Billieux, 2017, Young, 2008). Multiple terms have been used to describe the phenomenon (e.g., internet sex addiction, problematic online sexual activities, cybersex addiction, and problematic internet pornography use), and whether subjectively self-perceived addiction to pornography due to moral incongruence is regarded as PPU has been debated (Brand et al., 2019a; Vaillancourt‑Morel & Bergeron, 2019). Furthermore, there are no specific diagnostic criteria for PPU (Brand et al., 2020, Chen and Jiang, 2020, Cooper et al., 2001, Fernandez and Griffiths, 2019, Hertlein and Cravens, 2014, Wéry and Billieux, 2017). In order to study and treat PPU, researchers have developed scales that measure different aspects of PPU; however, few have been validated across cultures and different populations (Chen and Jiang, 2020, Fernandez and Griffiths, 2019, Wéry and Billieux, 2017).

2. Assessment of Problematic Pornography Use

Given debates on the conceptualization of and diagnostic criteria for PPU, assessment tools have varied across studies and emphasized different characteristics (Fernandez & Griffiths, 2019). Multiple scales have been based largely on proposed criteria for hypersexual disorder (e.g., the Hypersexual Behavior Inventory, Reid, Garos, & Fong, 2012). However, recent studies suggest differences between PPU and hypersexuality (Bőthe et al., 2019c). Hypersexuality may include high engagement in various sexual behaviors, including masturbation, cybersex, pornography use, telephone sex, sexual behavior with consenting adults, strip club visitations, and other behaviors (Karila et al., 2014). Consistently, the Hypersexual Behavior Inventory (HBI) assesses hypersexual behaviors more broadly (Brahim, Rothen, Bianchidemicheli, Courtois, & Khazaal, 2019). Some scales have focused on compulsive sexual behaviors more generally (e.g., Compulsive Use of Sexually Explicit Internet Material), with these scales assessing features of compulsive searching for/viewing of pornography on the internet (Doornwaard, Eijnden, Baams, Vanwesenbeeck, & Bogt, 2016), rather than those of general compulsive pornography use, and it did not undergo extensive psychometric evaluation. Some concise scales exist that aim to measure PPU, but these have, at times, been criticized or debated regarding their construct validity. For example, the Cyber-Pornography Use Inventory-9 (CPUI-9, Grubbs, Sessoms, Wheeler, & Volk, 2010) has been used to assess self-reported addiction and considers moral incongruence, although what it precisely measures have been questioned (Brand et al., 2019a). Several recent scales have been developed to assess aspects and domains of PPU more generally including the Short Internet Addiction Test Adapted to Online Sexual Activities (s-IAT-sex; Wéry, Burnay, Karila, & Billieux, 2015), the Problematic Pornography Use Scale (PPUS; Kor et al., 2014), and the Problematic Pornography Consumption Scale (PPCS-18; Bőthe et al., 2018b). The last two scales were recommended by a recent systematic review (Fernandez & Griffiths, 2019). More recently, compared with the PPUS and s-IAT-sex, the PPCS-18 demonstrated higher sensitivity and more accuracy in screening for PPU (Chen & Jiang, 2020).

The PPCS-18, to our knowledge, is the only instrument that assesses six specified components of one addiction model: salience, mood modification, conflict, tolerance, relapse, and withdrawal (Griffiths, 2005). In particular, tolerance and withdrawal are important dimensions of PPU that are not assessed by the PPUS and s-IAT-sex (Bőthe et al., 2018b; Fernandez & Griffiths, 2019). Compared with other measurements of PPU (i.e., the PPUS, s-IAT-sex, CPUI-9), another strength of the PPCS is that it is one of few instruments to provide a validated cutoff score (≥76, range 18-126) to differentiate problematic from non-problematic pornography use (Fernandez & Griffiths, 2019), which adds to its research and clinical utility. Another recently published screen, the Brief Pornography Screen (BPS, Kraus et al., 2020), also provides a cutoff (≥4, range 0-10) to screen for PPU. Given its brevity and its unidimensional structure, the BPS does not assess components such as tolerance. While cutoffs of usage time per week have been proposed (Cooper et al., 2000, Mechelmans et al., 2014), usage time is not consistently related to PPU (Bőthe, Tóth-Király, Potenza, Orosz, & Demetrovics, 2020b; Chen et al., 2019, Kühn and Gallinat, 2014). In addition, convergent and divergent validity of the PPCS has been supported in studies of sexuality-related (Bőthe, Tóth-Király, Demetrovics, & Orosz, 2017) and personality-related (Bőthe, Koós, Tóth-Király, Orosz, & Demetrovics, 2019a; Bőthe et al., 2019c; Bőthe, Tóth-Király, Potenza, Orosz, & Demetrovics, 2020b) variables.

Despite the previously presented strong psychometric properties of the PPCS-18, research is needed to further investigate its properties across cultural and clinical/subclinical contexts (Bőthe, Tóth-Király, Demetrovics & Orosz, 2020a; Bőthe et al., 2018b), as, for example, cultural characteristics may influence negative attitudes toward pornography use (Griffiths, 2012, Vaillancourt-Morel and Bergeron, 2019). It has been argued that pornography use may be self-considered as problematic in one cultural, religious, or moral background and possibly not in another (Grubbs & Perry, 2019). Prior PPCS-18 studies may have cultural limitations since they have been mainly conducted in Hungary (Bőthe et al., 2018a; Bőthe et al., 2019b; Bőthe et al., 2020a; Bőthe, Lonza et al., 2020). This may constitute a significant limitation since norms, value systems, and experiences of individuals from other cultural backgrounds may differ from largely Western perspectives in Hungary. Regarding pornography use and other sexual behaviors, differences in sexual attitudes, behaviors, and well-being have been reported between Eastern and Western cultures (Laumann et al., 2006). Thus, research on PPU is needed to ensure that assessments are both translatable and accurate across cultures (Kraus & Sweeney, 2019). There is relatively little empirical research on PPU in China and in other Eastern countries, and only several studies have included participants from Eastern countries (Fernandez & Griffiths, 2019), and cross-cultural community comparisons have not been examined.

Individuals with PPU may exhibit specific characteristics including strong cravings, poor self-control, continued engagement despite social or occupational impairments, and adverse consequences, and using pornography in maladaptive ways such as to escape from stress or negative mood states (Chen et al., 2018, Cooper et al., 2004, Kraus et al., 2016, Young et al., 2000). Wéry et al. (2016) reported that 90% of participants with PPU reported co-occurring psychiatric diagnoses, and only a few scales have been validated in treatment-seeking samples (Bőthe et al., 2020a; Kraus et al., 2020). Thus, in addition to the frequency of online sexual activities, craving, compulsive sexual behaviors, and general mental health were used to examine the criterion validity of the PPCS. In sum, primarily nonclinical and Western samples have been used in most studies of PPU assessments like the PPCS-18; therefore, more research is needed to validate the PPCS-18 across more diverse samples, including clinical or subclinical population and across cultures.

3. Network approach in psychopathology

Psychopathological states may exist as complex dynamic systems involving interacting components (Borsboom, 2017). In contrast to some latent models, network approaches propose that psychological disorders involve networks of related symptoms, and individual psychological states may rely more on direct connections between symptoms rather than on the existence of latent variables (Werner, Stulhofer, Waldorp, & Jurin, 2018). Network theories and methodologies have been fruitfully applied to different psychopathological phenomena including alcohol-use disorders (Anker et al., 2017), anxiety (Beard et al., 2016), depression (Schweren, van Borkulo, Fried, & Goodyer, 2018), and hypersexuality (Werner et al., 2018). Such network models may provide important insight into the centrality of specific domains and the patterns of their relationships. Therefore, in the current study, we used a network approach to assess PPU network topology and identify symptoms that occupy central positions in the network, and explored patterns of relationships of symptom domains in the different populations. This approach will provide insight into how PPU may interact with symptomatology across cultures and community and subclinical samples.

4. The aims of the current study

Considering that men relative to women typically show stronger cravings for pornography and more frequent use (Weinstein, Zolek, Babkin, Cohen, & Lejoyeux, 2015), more frequent PPU (Kafka, 2010, Kraus et al., 2016, Kraus et al., 2015), and more treatment-seeking for PPU (Bőthe et al., 2020a), the aims of the present study were (1) to examine the reliability, structure and convergent validity of the PPCS-18 both in community and subclinical samples of Chinese men; and (2) to examine and compare the factor structure of PPCS-18 across Hungarian and Chinese samples, and across community and subclinical samples; and, (3) to explore the extent to which the PPCS-18 reflects characteristics related to the different populations in network typology analyses.

5. Method

6. Measures

Brief Pornography Screen (BPS, Kraus et al., 2020)¹. The BPS is a screening tool for PPU (Efrati and Gola, 2018, Gola et al., 2017). It is a five-item assessment and uses a three-point rating scale for each item (0 = never, 1 = occasionally, 2 = always). The Cronbach’s alpha of the BPS was .89 in the Chinese community sample and .74 in the Chinese subclinical sample.

Pornography Craving questionnaire (PCQ, Kraus & Rosenberg, 2014). This 12-item questionnaire is a unidimensional assessment (Kraus and Rosenberg, 2014, Rosenberg and Kraus, 2014). The respondents were required to indicate how strongly they agreed with each item using the following seven response options (presented without numerals): “completely disagree,” “somewhat disagree,” “disagree a little,” “neither agree nor disagree,” “agree a little,” “somewhat agree” and “completely agree.” Higher scores are indicative of a greater craving for pornography. The Chinese version of the PCQ has been used in a previous study (Chen et al., 2019). The Cronbach’s alpha of this scale was .92 in the Chinese community sample and .91 in the Chinese subclinical sample.

Sexual Compulsivity Scale (SCS, Kalichman & Rompa, 1995). The extent to which participants exhibit characteristics of sexual compulsivity was assessed using the ten-item SCS. Responses were recorded on a four-point rating scale (1 = not at all like me, 2 = slightly like me, 3 = mainly like me, 4 = very much like me). The Chinese version of the SCS has been previously described (Chen & Jiang, 2020). The SCS demonstrated excellent reliability in the present study (α was .91 in community men and .90 in subclinical men).

Questionnaire of online sexual activities Chinese version (OSAs, Zheng & Zheng, 2014). Thirteen items were used to measure participants’ use of the internet for the following purposes: (1) viewing sexually explicit material (SEM), (2) seeking sexual partners, (3) cybersex, and (4) flirting and sexual relationship maintenance. The Cronbach’s alpha of the entire scale was .84 in the Chinese community men and .81 in subclinical men. Higher scores were indicative of more frequent engagement in OSAs.

12-item General Health Questionnaire (GHQ-12, Goldberg & Hillier, 1979). The GHQ-12 is a widely used screening instrument for common mental disorders and is recommended as a case detector since it is considered brief, effective and robust and works as well as its longer versions (Goldberg et al., 1997, Petkovska et al., 2015). The GHQ-12 has been translated into many languages, including Chinese, and its psychometric properties have been studied among many different populations (Pan and Goldberg, 1990, Petkovska et al., 2015). The GHQ-12 includes 12 items total (six positive and six negative), each scored on a four-point Likert scale, with higher scores reflecting worse psychological health. The Cronbach’s alpha of the scale was .89 in Chinese community men and .93 in subclinical men.

7. Statistical Analyses

First, CFA was conducted on the Hungarian men, then on Sample 1 and Sample 2 to cross-validate the results in the community and subclinical samples of Chinese men. The mean- and variance-adjusted weighted least squares estimator (WLSMV) was used for parameter estimation. Model fit indices were determined by Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), and the Root Mean Square Error of Approximation (RMSEA) and standardized root mean square residual (SRMR). CFI and TLI values greater than .95 were considered as excellent fit (≥.90 for acceptable fit). RMSEA values of less than .06 were considered excellent (≤ .08 for adequate fit, and ≤ .10 for acceptable fit with its 90% confidence interval) (Browne and Cudeck, 1993, Schermelleh-Engel et al., 2003). SRMR values less than 0.08 (≤.06 for a good fit) were considered indicative of an acceptable model (Hu & Bentler, 1999). Additionally, to test measurement invariance among different cultural contexts (Hungarian and Chinese), and community and subclinical populations, multi-group CFAs were conducted on the three samples. Six levels of invariance were tested and compared in each case: configural, metric, scalar, residual, latent variance, and latent mean. When comparing the increasingly constrained models, relative changes in fit indices were observed, with a recommended acceptable range as follows: ΔCFI ≤ .010; ΔTLI ≤ .010; and ΔRMSEA ≤ .015 (Meade, Johnson, & Braddy, 2008).

Cronbach’s alpha and Composite Reliability (CR) values were also calculated. Associations between the sexual compulsivity scale (SCS), pornography craving questionnaire (PCQ), general health questionnaire (GHQ-12), frequency of OSAs, BPS, and PPCS-18 were assessed to corroborate the validity of the PPCS-18. Correlations between variables were examined using Pearson correlation coefficients after controlling for age, sexual orientation, and relationship status.

We estimated and analyzed the PPCS-18 networks in two steps. The first step was to establish a regularized network, also known as a markov random field. LASSO regression was adopted for adjustment to reduce the appearance of false connections. As described previously (Epskamp & Fried, 2017), the EBIC hyperparameter was set at .5. Second, we assessed the relative position of nodes using centrality statistics and tested three common centrality metrics: node strength, closeness, and betweenness centrality. Among them, betweenness centrality refers to the number of times that a node stays on the shortest path between other nodes. Closeness centrality is the inverse of the sum of the shortest paths from one node to all other nodes. In addition, we compared the global strength of connectivity for each network (i.e., the sum of all the associated strengths) using the Network Comparison Test. All network analyses were performed using the qgraph, dplyr, NetworkComparisonTest, and bootnet packages in R. (Version 3.6.2).

8. Results

9. Measurement invariance test of the PPCS-18 across cultures and in community and subclinical men

The results of measurement invariance are shown in table 3. For configural invariance, RMSEA was slightly higher than the recommended threshold value (i.e., .10), but the model demonstrated acceptable fit indices on the CFI, TLI, and SRMR. Thus, we retained this model for the further steps of invariance testing. In the metric model, the fit indices were more suitable compared to the preceding model. Then, scalar and residual invariance were achieved, but latent mean invariance was not, suggesting the presence of latent mean differences between the community and subclinical men (see Table 3). When the subclinical men’s latent mean differences were constrained to zero for the purpose of model identification, individuals’ latent means in the community men were substantially lower than participants’ latent means in the subclinical men (Sample 1: -0.88 to -1.81 SD in the six factors, p < .001; Sample 3: -0.39 to -2.46 SD in the six factors, p < .01), indicating that subclinical individuals demonstrated significantly higher scores on the PPCS than those in the Chinese and Hungarian community samples. In sum, the PPCS-18 had similar meanings and latent structure in Chinese and Hungarian community men, and it may be used in comparisons of Chinese and Hungarian men.

10. Interaction of the six factors of the PPCS-18 in each sample

Results of Markov random fields showed that there was a significant difference between the Hungarian and Chinese men (p < .01). Among the Chinese community and subclinical men, conflict was negatively related to salience; otherwise, conflict did not relate to salience directly, and had positive correlations with other factors among Hungarian men (see Figure 1). Schematic diagrams of Chinese community and subclinical men were similar, and no significant difference in the global strength of connectivity was observed (p = 0.6). Centrality estimates are presented in Figure 2 (centrality plots). In the three samples’ networks, withdrawal was the most central node, while tolerance was also a central node in the subclinical individuals’ network. In support of these estimates, withdrawal was characterized by high predictability in all networks (Chinese community men:76.8%, Chinese subclinical men: 68.8%, and Hungarian community men: 64.2%).

1. Download : Download high-res image (92KB)
2. Download : Download full-size image

Figure 1. Network schematic diagram in three groups of men. Notes. The Chinese community men’s network is presented on the left and the Hungarian community men’s network on the right. The middle is the network of Chinese subclinical sample men. Solid edges indicate positive and dashed edges indicate negative relationships.

1. Download : Download high-res image (214KB)
2. Download : Download full-size image

Figure 2. Node centrality plot in the three groups of men

11. Discussion

Although several scales for assessing PPU are available to researchers and clinicians, few have been subsequently revalidated across different cultures, and the psychometric properties of scales in subclinical men have rarely been examined. Additionally, how symptom domains related to PPU relate (i.e., interrelationships between salience, withdrawal, tolerance, mood modification, conflict, and relapse) in such samples is poorly understood (Bőthe, Lonza, et al., 2020). Therefore, we examined the reliability and validity of the PPCS-18 in Chinese contexts and demonstrated support for its use in Chinese community and subclinical men. The Chinese version of the PPCS-18 demonstrated high internal consistency, composite reliability, and convergent validity in both Chinese community and subclinical men. Measurement invariance testing suggested that the scale was similarly applicable to the Hungarian community, the Chinese community, and the Chinese subclinical populations, supporting the scale’s potential cross-cultural and clinical utility. Network analysis showed that the interaction between the six factors of the PPCS-18 was significantly different in Hungarian and Chinese men. Centrality estimates indicated that the subclinical sample’s core symptoms were withdrawal and tolerance, but only the withdrawal domain was a central node in both community samples.

12. Validity and reliability of the PPCS-18 in Chinese populations

The construct validity and reliability of the PPCS-18 were cross-validated on these three independent and distinct samples. Not only was the construct validity of the PPCS-18 supported, but also its convergent validity was established by reporting its associations with pornography craving, compulsive sexual behaviors, frequencies of OSAs, and participants’ general psychological health levels. Similar to a prior study (Bőthe et al., 2020b), the frequency of OSAs did not appear as reliable an indicator of PPU, due to the correlation coefficients between four sub-types of OSAs and PPCS-18 ranging from small to large, which suggests that the PPCS-18 may also be sensitive to quantitative aspects of PPU in Chinese contexts, although this possibility warrants additional study.

Besides the frequency of consumption, qualitative aspects such as content that may elicit pornography craving should be considered (Kraus & Rosenberg, 2014). The subjective experience of craving is a common element of addictions (Kraus & Rosenberg, 2014), and is relevant in predicting the occurrence, maintenance and relapse of addictive behaviors after withdrawal (Drummond, Litten, Lowman, & Hunt, 2000). Consistent with previous studies (Gola and Potenza, 2016, Young et al., 2000), worse mental health scores and more compulsive sexual behaviors correlated with higher PPCS scores. These results suggest it may be advisable to consider craving, mental health factors, and compulsive use in screening and diagnosing PPU (Brand, Rumpf et al., 2020).

13. Networks of PPU symptoms in community and subclinical men

Similar to the application of a network approach in hypersexuality (Werner et al., 2018), we applied this approach to PPU in order to examine whether the PPCS-18 demonstrates similar or distinct relationships across different samples. The overall network topologies of the three samples suggest that the relationships between domains of the PPCS-18 may have culture-related differences. In Chinese men, the conflict factor was negatively associated with salience, while in Hungarian men, salience did not relate to conflict. In parallel with massive socio-economic changes over past decades in China, increasingly more Chinese people are criticizing conservative sexual attitudes, especially those who define sex as immoral, and instead, they have begun to highlight the importance of sexual pleasure (Lin, 2018, Wong, 2014). In the current study, participants were men. In predominating sexual scripts in China, men are encouraged to pursue sexual expression and exhibit more permissive sexual attitudes (Zheng et al., 2011). Therefore, when men’s thoughts may be focused on pornography, they may not experience conflict. On the other hand, the assessment of the “conflict” component on the PPCS is limited to its inclusion of more peripheral aspects of conflict (e.g., negative effects on sexual life) and exclusion of more central aspects of conflict (e.g., interpersonal conflict) (Fernandez & Griffiths, 2019). However, the precise reasons for underlying differences in relationships between Chinese and Hungarian men in relationships between conflict and salience warrant additional study, particularly given that factors like social acceptance and governmental regulation of pornography use may differ across jurisdictions.

Additionally, centrality estimates in the six factors of the PPCS-18 displayed withdrawal as the most crucial factor in all three samples. According to the strength, closeness, and betweenness centrality results among subclinical participants, tolerance also contributed importantly, being second only to withdrawal. These findings suggest that withdrawal and tolerance are particularly important in subclinical individuals. Tolerance and withdrawal are considered as physiological criteria relating to addictions (Himmelsbach, 1941). Concepts like tolerance and withdrawal should constitute a crucial part of future research in PPU (de Alarcón et al., 2019, Fernandez and Griffiths, 2019). Griffiths (2005) postulated that tolerance and withdrawal symptoms should be present for any behavior to be considered addictive. Our analyses support the notion that withdrawal and tolerance domains are important clinically for PPU. Consistent with Reid’s view (Reid, 2016), evidence of tolerance and withdrawal in patients with compulsive sexual behaviors may be an important consideration in characterizing dysfunctional sexual behaviors as addictive.

14. Limitations and future studies

The present study is not without limitations. First, temporal stability was not tested. Second, data were collected using self-report measures; therefore, the reliability of the results depends on respondents’ honesty and accuracy and their comprehension of items. Third, the RMSEA value was slightly higher in the subclinical samples, warranting further research. Participants included only men aged 18-48 years; thus, the applicability of the PPCS-18 in older populations and women should be further examined. It still unclear whether gender-related differences may be influenced by cultural or jurisdictional factors. Hence, more research is needed to validate the PPCS-18 across more diverse samples, including women, diverse age groups, and other cultures and jurisdictions. Additionally, the subclinical group studied was derived from an online forum. The extent to which the findings may extend to other settings (e.g., those providing face-to-face treatment) warrants further study.

15. Conclusions

The PPCS-18 had strong psychometric properties in community men from Hungary and China, and subclinical men from China who reported poorly controlled pornography use. Thus, the PPCS-18 appears to be a valid and reliable measure to assess PPU across specific Western and Eastern jurisdictions and may be used among subclinical individuals. Furthermore, relationships between the PPCS-18 domains also can reflect distinct characteristics of different populations, and the current findings suggest that withdrawal and tolerance are important to consider in PPU. The findings advance understanding by reporting subclinical and community samples in China, expanding the generalizability of the PPCS-18, and exploring relationships between different symptom domains across cultures.

Funding

The research was supported by the National Social Science Foundation of China (Grant No. 19BSH117 and CEA150173) and the Education Reform Project of Fujian province (FBJG20170038). BB was funded by a postdoctoral fellowship award by Team SCOUP – Sexuality and Couples – Fonds de recherche du Québec, Société et Culture. ZD was supported by the Hungarian National Research, Development and Innovation Office (Grant numbers: KKP126835, NKFIH-1157-8/2019-DT). MNP’s involvement was supported by the National Center for Responsible Gaming through a Center of Excellence grant. The funding agencies did not have input into the content of the manuscript and the views described in the manuscript reflect those of the authors and not necessarily those of the funding agencies.

Conflict of interest

The authors declare no conflict of interest with respect to the content of this manuscript.

Uncited references

Bőthe et al., 2018, Bőthe et al., 2019, Bőthe et al., 2019, Bőthe et al., in press, Bőthe et al., 2020, Bőthe et al., 2019, Bőthe et al., 2020, Bőthe et al., 2018, Brand et al., 2019, Brand et al., 2019.