The Association between the Quantity and Severity of Pornography Use: A Meta-analysis

ABSTRACT

Although the quantity of pornography use (QPU, i.e., frequency/time spent on pornography use) has been positively associated with the severity of pornography use (i.e., problematic pornography use, PPU), the magnitudes of relationships have varied across studies. This meta-analysis aimed to assess the overall relationships and identify potential moderating variables to explain the variation in these associations between QPU and PPU. We performed a literature search for all published and unpublished studies from 1995 to 2020 in major online scientific databases up until December 2020. Sixty-one studies were identified with 82 independent samples involving 74,880 participants. Results indicated that there was a positive, moderate relationship between QPU and PPU (r = 0.34, p < .001). The strength of relationship significantly varied across measures of PPU based on different theoretical frameworks, indicators of QPU, and sexual cultural contexts (conservative vs. permissive sexual values). Frequency was a more robust quantitative indicator of PPU than time spent on pornography use. In conservative countries, QPU showed more robust association with self-perceived PPU. Future studies are encouraged to select the measurement of PPU according to research aims and use multi-item measures with demonstrated content validity to assess pornography use. Cross-cultural (conservative/permissive) comparisons also warrant further research.

Over the past two decades, a revolution in information and communication technology has altered sexual behaviors, including pornography use. The conceptual meaning of “pornography use” is broad, as there are several different ways that one can define pornography and consider its use (Busby et al., 2020; Kohut et al., 2020). Recently, the concept of pornography use is suggested to be defined as “a common but stigmatized behavior, in which one or more people intentionally expose themselves to representations of nudity which may or may not include depictions of sexual behavior, or who seek out, create, modify, exchange, or store such materials. Pornography use can evoke immediate sexual and affective responses, and may contribute to more lasting cognitive, affective, and behavioral changes” (Kohut et al., 2020, p. 12).

Two major aspects of pornography use including the quantity and quality/severity of use are present in the literature when examining pornography use (Gola et al., 2016). The frequency and amount of time spent using pornography are most often considered as the quantity of pornography use (QPU) (Kohut et al., 2020). Yet, it needs to be noted that pornography use is not necessarily associated with significant impairments in different domains of functioning. However, pornography use may become problematic (Blais-Lecours et al., 2016; Engel et al., 2019; Harper & Hodgins, 2016). The quality/severity of pornography use (PPU) usually represents the extent of pornography negatively impacting people’s lives and dominating their thinking, feelings, and behaviors (Bőthe, Tóth-Király, Griffiths et al., 2021; Chen, Jiang et al., 2021). Thus, most researchers consider PPU as persistent, repetitive engagement in pornography use that results in impairment in one’s life in addition to failed attempts to reduce or stop such behaviors (Bőthe, Tóth-Király, Demetrovics et al., 2021; Chen, Yang et al., 2018; Cooper et al., 2004; Efrati & Gola, 2018; Kraus, Martino et al., 2016; Young et al., 2000). Quantity of pornography (QPU) was identified as a peripheral symptom (i.e., symptoms that might not belong to the core symptoms of problematic behaviors and might be present in the case of non-problematic, highly engaged users) in PPU (Bőthe, Lonza et al., 2020), and it may not be a sufficient indicator of PPU (Bőthe, Lonza et al., 2020; Bőthe, Tóth-Király et al., 2020; Vaillancourt-Morel et al., 2017). Yet, it can reflect the degree of PPU to some extent (Chen, Ding et al., 2018), as previous studies have found that QPU has been positively associated with PPU. However, the magnitudes of these relationships have varied across studies.

Cross-sectional studies have found weak-to-large relationships between frequencies or usage times of pornography use and PPU in undergraduate or adult samples (Bőthe, Tóth-Király et al., 2018; Chen, Ding et al., 2018; Fernandez et al., 2017; Gola et al., 2016; Grubbs, Exline et al., 2015; Grubbs et al., 2017; Grubbs & Gola, 2019; Grubbs, Kraus et al., 2019; Grubbs, Wilt, Exline & Pargament, 2018; Volk et al., 2016; Wéry et al., 2016). Recent longitudinal studies have investigated these relationships and found that QPU demonstrated small correlations with self-perceived PPU (Grubbs & Gola, 2019; Grubbs, Stauner et al., 2015; Grubbs, Wilt, Exline, Pargament et al., 2018). These findings to date indicate that complex associations might exist between QPU and PPU, suggesting a need for a more comprehensive understanding. Further, identification and analysis of the moderators from a sufficient number of studies can add valuable knowledge to our understanding of how the links between QPU and PPU vary by study characteristics; thus, the present meta-analysis might help identify potential knowledge gaps in the literature and areas to target in future research.

If we wish to better understand the associations between QPU and PPU, both measurement characteristics (QPU and PPU), and individual characteristics should be considered as potential moderators. For instance, QPU assessed with a single item (Kohut & Stulhofer, 2018; Rousseau et al., 2021) or multiple items (Hatch et al., 2020; Leonhardt et al., 2021) and PPU measured by scales based on different theoretical frameworks (Chen & Jiang, 2020; Fernandez & Griffiths, 2021) may have an effect on the results. Individual characteristics may moderate these associations as well. For example, gender (Pekal et al., 2018; Weinstein, Zolek et al., 2015), sexual cultural context (Chen, Luo et al., 2021; Griffiths, 2012; Kraus & Sweeney, 2019; Vaillancourt-Morel & Bergeron, 2019), sexual orientation (Brand, Brand, Antons et al., 2019; Duggan & McCreary, 2004; Træen & Daneback, 2013) and treatment-seeking vs. community groups (Bőthe, Tóth-Király, Demetrovics et al., 2021; Bőthe, Tóth-Király et al., 2018; Klucken et al., 2016; Lewczuk et al., 2017; Voon et al., 2014) may have a direct or interaction effect on QPU or PPU.

Literature Review and Potential Moderators

Measurement Characteristics

Assessments of PPU based on different theoretical frameworks. Currently, there is no consensus regarding the conceptualization and diagnosis of PPU (Grubbs et al., 2020). Some researchers have considered conceptualization of PPU as an impulse control or compulsivity-related behavior problem (Kuhn & Gallinat, 2014; Meerkerk et al., 2009; Noor et al., 2014), or a potentially addictive behavior (Kraus, Voon et al., 2016). However, in some researchers’ opinion, PPU is not yet recognized as a psychiatric disorder because PPU may not necessarily be excessive, addictive, or compulsive (Vaillancourt-Morel & Bergeron, 2019), and there are individuals who experience self-perceived PPU due to moral incongruence (Grubbs, Perry et al., 2019).

Impaired control (i.e., unsuccessful attempts to reduce or stop a given behavior and difficulties in controlling related urges) is an important feature of impulse control, compulsivity-related, and addictive behaviors as well (Fineberg et al., 2014). However, important differences can be observed regarding the nature of impaired control in the different disorders. While impaired control may appear in relation to risk-seeking and risk-taking in the case of impulse control disorders, impaired control is associated with repetitive engagement in behaviors in a habitual manner according to rigid rules to avoid harm in compulsive disorders. In the case of addiction models, both impulsivity and compulsivity-related impaired control may be present (Brand, Wegmann et al., 2019; Brand et al., 2016; Fineberg et al., 2014). Moreover, impaired control may also be an essential criterion to distinguish between individuals with subjective, self-perceived PPU and dysregulated PPU (Chen, Jiang et al., 2021). Lastly, addiction models may assess tolerance and withdrawal, two features of addictions (Griffiths, 2005), that are pivotal characteristics distinguishing between addictive disorders and obsessive-compulsive and impulse control disorders (Reid, 2016). In sum, scales using different theoretical frameworks to assess PPU may focus on different elements or characteristics of impaired control in PPU and may include different symptoms (e.g., tolerance). Therefore, considering the potential role of using scales developed on the basis of different theoretical frameworks is crucial.

Corresponding with the debate on the conceptualization of PPU, over 20 psychometric instruments have been developed to assess various conceptualizations and aspects of PPU (Fernandez & Griffiths, 2021; Karila et al., 2014). Associations between QPU and PPU have varied considerably when different PPU measures have been used. Some scales assessing PPU arguably based on obsessive-compulsive frameworks have demonstrated moderate associations with QPU (e.g., Brief Pornography Screen; Sexual Compulsivity Scale) (Lewczuk et al., 2020; Wetterneck et al., 2012), while scales developed on addiction models (e.g., Griffiths, 2005) showed small-to-large associations between QPU and PPU [e.g., Compulsive Internet Use Scale adapted to sexual explicit materials (CIUS-adapted sexual explicit materials) (Brahim et al., 2019), and Problematic Pornography Use Scale (PPUS) (Chen, Yang et al., 2018; Kor et al., 2014]. However, some researchers developed scales to underscore individuals’ subjectively self-perceived addictions [i.e., the Cyber Pornography Use Inventory, (Grubbs et al., 2010; Grubbs, Volk et al., 2015)], and demonstrated moderate associations between the QPU and PPU. These findings suggest that differences in measurements of PPU may be an important variable explaining variability of strengths in associations between QPU and PPU.

Measurement of QPU. Measures of QPU have primarily included frequency and usage time of pornography use during a given period of time (e.g., past week). Previous studies have suggested that important differences in associations may be present when these two measures of QPU are assessed (Chen, Ding et al., 2018; Fernandez et al., 2017). Based on prior studies, PPU may be weakly related to the time spent viewing pornography per occasion, while it might have a stronger association with the frequency of use (Bőthe, Tóth-Király et al., 2018; Chen & Jiang, 2020; Grubbs, Kraus et al., 2019; Rimington, 2008). However, other studies reported that usage time had a stronger association with PPU than frequency of use (Antons et al., 2019). To date, no solid conclusions may be drawn from previous studies whether the frequency of pornography use or the usage time of pornography has a stronger association with PPU.

In reviewing the nature of the QPU scales, most researchers have used a single general question to ask participants to estimate their average QPU (Blais-Lecours et al., 2016; Bőthe, Tóth-Király et al., 2020; Bradley et al., 2016; Fernandez et al., 2017; Wetterneck et al., 2012; Wilt et al., 2016), while few studies have used self-compiled multi-item assessments (Bőthe, Bartók et al., 2018; Harper & Hodgins, 2016; Wéry et al., 2016) or validated scales (Brahim et al., 2019; Chen & Jiang, 2020; Chen, Yang et al., 2018) such as the Online Sexual Activities questionnaire [OSAs, including a five-item questionnaire for measuring the viewing sexual explicit materials, (L. Zheng & Zheng, 2014)]. Whether these two measurement methods have an impact on the strengths of the associations between QPU and PPU has remained unclear and will be examined in the current study.

Sample Characteristics

Gender. Considerable differences may exist in the relationships between QPU and PPU in men and women. Compared to women, men typically consume more pornography as measured by usage time and frequency (Hald, 2006; Martyniuk et al., 2016; Morgan, 2011), and men on average experience higher levels of PPU than women (Bőthe, Tóth-Király et al., 2018; Grubbs, Kraus et al., 2019). Concerning associations between QPU and PPU (Gola et al., 2016; Lewczuk et al., 2017), moderate-to-strong relationships were reported between usage time and self-perceived PPU among women (Baranowski et al., 2019; Laier et al., 2014), while weak-to-strong correlations were reported among men (Antons et al., 2019; Laier et al., 2015; Wéry et al., 2016). Moreover, previous research has demonstrated that the difference in pornography use is one of the greatest differences in sexual behavior between men and women (Petersen & Hyde, 2010). However, questions on whether the QPU mirror the severity of pornography use to the same extent between genders have not been explored yet.

Cultural context toward sexual attitudes (conservative/ permissive). The sociocultural context may play an important role in influencing people’s attitudes toward sexual behaviors (Griffiths, 2012; Meston et al., 1996) and females’ sexual expression (Guo, 2019; Petersen & Hyde, 2010). Furthermore, cultural characteristics may influence negative attitudes toward pornography use (Vaillancourt-Morel & Bergeron, 2019), and some have proposed that conservative sexual cultures may influence self-perception of PPU as well (Grubbs, Wilt, Exline & Pargament, 2018). Individuals in sexually conservative cultures [e.g., traditional Asian cultures that stress sexual conservatism (Kim, 2009)] may feel more conflict and negative emotions about their use given that it may be in opposition to their sexual moral values, and may be more likely to label themselves as being addicted to pornography. For example, associations between QPU and PPU were stronger in a Chinese sample, compared to a US sample using the same PPU scale (r = 0.60 vs 0.39, respectively) (Chen, Yang et al., 2018; Shirk et al., 2021). Furthermore, although conservative cultures stipulate sexual restriction for both men and women, female sexuality is typically more restricted, especially for girls and unmarried women (Browning et al., 2006; Kim, 2009; Okazaki, 2002). Therefore, there may be interactions between gender and sexual culture context. For example, while QPU may have a significant impact on women’s PPU levels in more conservative cultures, men’s use might not be as limited by cultural expectations and thus they may not perceive their use as problematic. However, whether the associations between QPU and PPU vary across different sexual culture contexts, and whether there is an interaction between gender and sexual cultural context, have not been examined yet.

Sexual orientation. As a core personal characteristic relevant to sexual behaviors, individuals’ sexual orientation may be considered as a potential variable when examining associations between QPU and PPU (Brand, Brand, Antons et al., 2019). Sexual minority men or women typically use more pornography than their heterosexual counterparts (Duggan & McCreary, 2004; Træen & Daneback, 2013; Træen et al., 2006). Sexual minority individuals may also report higher levels of sexual compulsivity compared to heterosexual individuals (Bőthe, Bartók et al., 2018; Cooper et al., 2000; Weinstein, Katz et al., 2015). A possible explanation may be that sexual minorities may experience more negative effects (such as anxiety, depression, anger or stress) in relation to their sexual orientation (Parsons et al., 2008), and online pornography, given its accessibility, affordability and perceived anonymity (Cooper, 1998), may be a commonly used coping strategy to reduce such unwanted feelings (Bőthe, Bartók et al., 2018; Parsons et al., 2008).

Community and treatment-seeking participants. Although multiple scales exist for assessing PPU, primarily nonclinical samples have been examined in most studies examining PPU (Fernandez & Griffiths, 2021). Participants with PPU typically report higher QPU relative to healthy volunteers (Gola et al., 2017; Prause et al., 2015; Voon et al., 2014), and typically there exist small-to-moderate associations between QPU and PPU in community samples (Bőthe, Tóth-Király et al., 2018; Brand et al., 2011; Grubbs, Exline et al., 2015; Leonhardt et al., 2018; Twohig et al., 2009; Wetterneck et al., 2012), while stronger, moderate associations have been reported in treatment-seeking and clinical samples (Gola et al., 2016; Lewczuk et al., 2017). Still, other findings suggest that associations might be stronger in community samples than in treatment-seeking samples (Bőthe, Lonza et al., 2020). Thus, clinical/treatment-seeking status is important to consider in understanding associations between QPU and PPU.

Aims of the Present Study

Given variability in strengths of associations between QPU and PPU in different studies, the present study aimed to use meta-analytic methods to evaluate the overall association between QPU and PPU and investigate potential moderators, such as sociodemographic, cultural, and measurement-related characteristics in these associations. Specifically, we explored which measurement characteristics (e.g., assessments of PPU and QPU) and sample characteristics (e.g., gender, sexual orientation, sexual cultural context and sample type) may explain the inconsistent associations between QPU and PPU.

Method

Literature Search

Major electronic databases, including Web of Science, PubMed, ProQuest, Science Direct and Wiley Online Library were searched in November 2019 (first time), April 2020 (second time), August 2020 and (third time), December 2020 (fourth time) for the following search terms: (porn* OR “sexually explicit” OR erotica OR “adult films” OR “adult movies” OR “adult videos” OR “x-rated films” OR “x-rated movies” OR “x-rated videos”) and (problem* OR compul* OR impul* OR addict* OR excess* or dependen*). We searched for these terms in the keywords, title, and abstract sections. The year 1995 was chosen as an initial time for articles given that since the late 1990s the conceptualization of PPU has been proposed and the internet has changed access to pornography (Grubbs et al., 2020).

Selection Criteria

Studies included in this meta-analysis required that they (a) reported original empirical findings; (b) included adult or adolescent samples (if there were multiple reports of the same study sample, the first published report was used); (c) reported associations between QPU (usage time or frequency of PU) and PPU (total score or at least one dimension of PPU)¹; (d) focused on PPU (articles focusing on other topics such as paraphilia, sexual abuse, sex crimes or hypersexuality were not retained)²; (e) were published in English or Chinese (the languages spoken by the authors). We contacted the papers’ authors to request correlation coefficients if they were not reported in the published article (see Figure 1 for the flow diagram of the article-sele tion process).

Figure 1. Flow diagram of the article selection process.

Coding and Quality Evaluation

Two authors independently screened the titles and abstracts of the identified manuscripts to select articles; then, full-text manuscripts were assessed for eligibility. Disagreements were resolved by discussion. Studies that met inclusion criteria were coded for the following information: first author’s name, year of publication, country, sample size and sample characteristics (i.e., age, gender, sexual orientation, conservative/permissive sexual cultural context, and sample type), measurement of PPU, and measurement of QPU (i.e., frequency or usage time). The classification of conservative/permissive sexual cultural context was based on the sexual value results from The World Values Survey³ (WVS, for more details see Supplementary Material 1, Table S1) (Haerpfer et al., 2020). The WVS asks about a number of issues related to tolerance toward divorce, abortion, and homosexuality. A low acceptance rate (i.e., lower scores than the average on these items in the WVS) of these issues are considered reflecting conservative values (W. Zheng et al., 2011). In the present study, China, Malaysia, and Poland were included in the conservative group, while the permissive countries group included Hungary, Croatia, Canada, Israel, United States, Switzerland, Czech Republic, Germany, Spain, and Australia.

Quality of included studies was assessed by a modified version of the quality appraisal tool recommended by Crombie (1996) for assessment of cross-sectional studies. There are seven criteria to rate the quality of the included studies. These criteria covered “Appropriate research design and statistical analysis,”⁴ “Appropriate data collection strategy,”⁵ “Inclusion and exclusion criteria for participation,” “Representativeness of samples (age, diversity),” “Reliability of PPU measure (Cronbach’s α in present study),”⁶ “Description of QPU measure,” and “Publication status (published or unpublished).” More details about the quality of the included studies are available in Supplementary Material 2, Table S2). The total quality scores ranged from 0 to 7 (M = 6.14, SD = 0.62). The inter-rater reliability of the quality evaluation score was high (r = 0.94). Disagreements were resolved by discussion. Table 1 summarizes the major characteristics of the included studies.

Table 1. Characteristics of the 61 studies included in the present meta-analysis.

Studies	PPU scales	QPU	Sample size (Male/Female)	Age (in years) M(SD); Range	Country	Sexual attitudes of countries based on the World Values Survey dataset	Sexual orientation (ratio of heterosexual participants)	Quality evaluation ^a (range 0–7)	Effect size (r)
Antons et al. (2019)	s-IAT-sex	frequency	1498/0	31.76 (11.26); 18–83	Germany	Permissive	1.00	6.5	0.219
		usage time							0.286
Ballester-Arnal et al. (2017)	ISST-Compulsivity	usage time	554/1003	20.37 (1.98); 18–25	Spain	Permissive	0.94	4.5	0.494
Baranowski et al. (2019)	s-IAT-sex	usage time	0/482	25.79 (7.27); 18–77	Germany	Permissive	-	6.5	0.430
Blais-Lecours et al. (2016)	CPUI-9	usage time	235/597	25.20 (7.99); 18–78	Canada	Permissive	-	6	0.343
	SCS								0.300
Borgogna, Isacco et al. (2020)	CPUI-9	frequency	244/0	19.63 (6.22); -	USA	Permissive	1.00	6	0.120
Borgogna et al. (2018)	PPUS	frequency	0/949	22.80 (8.50); 18-	USA	Permissive	0.82	5.5	0.453
Borgogna, McDermott et al. (2020)	PPUS	frequency	520/0	25.46 (11.99); 18–80	USA	Permissive	0.83	6.5	0.410
Bőthe, Lonza et al. (2020)	PPCS-6	frequency	509/0	34.13 (9.65); 18–76	Hungary	Conservative	-	7	0.383
	PPCS-6	frequency	3684/0	38.92 (12.61); 18–76	Hungary	Conservative	-	7	0.444
Bőthe, Tóth-Király, Bella et al. (2021)	PPCS	frequency	545/537	24.3 (5.4); 18–64	Hungary	Conservative	0.82	7	0.220
Bőthe, Tóth-Király, Demetrovics et al. (2021)	PPCS	frequency	258/8	37.2 (12.3); 18–76	Hungary	Conservative	0.69	6.5	0.270
		usage time							0.230
	PPCS	frequency	10,581/4470	33.2 (11.0); 18–76	Hungary	Conservative	0.83	6.5	0.510
		usage time							0.300
	PPCS	frequency	712/48	39.6 (9.7); 19–75	Hungary	Conservative	0.79	6.5	0.300
		usage time							0.250
Bőthe, Tóth-Király et al.(2020)	PPCS	frequency	231/252	27.7 (9.3); 18–73	Hungary	Conservative	0.90	7	0.460
Bőthe, Tóth-Király et al. (2018)	PPCS	frequency	382/390	22.58 (4.89); 18–54	Hungary	Conservative	0.80	7	0.290
		usage time							0.140
Bradley et al. (2016)	CPUI-9	usage time	370/338(5 not mentioned)	30.20 (9.90); -	USA	Permissive	-	6.5	0.180
Brahim et al. (2019)	CIUS-adapted sexually explicit material	frequency	150/156	32.63 (10.83); 18–69	Switzerland	Permissive	0.84	6.5	0.160
Brand et al. (2011)	IAT-sex	frequency	89/0	23.98 (4.09); -	Germany	Permissive	1.00	5.5	0.365
		usage time							0.203
Brand et al. (2016)	s-IAT-sex	frequency	19/0	25.05 (1.43); -	Germany	Permissive	1.00	5	0.290
		usage time							0.400
Chelsen (2011)	CPUI-9	usage time	530/0	undergraduate students	USA	Permissive	-	5	0.620
Chen, Ding et al. (2018)	s-IAT-sex	frequency	622/448	20.19 (1.18); 18-	China	Conservative	-	6.5	0.350
		usage time							0.280
Chen and Jiang (2020)	PPCS	frequency	192/125	24.64 (7.39); 18–48	China	Conservative	0.92	7	0.670
		usage time							0.280
	PPUS	frequency	201/131	24.47 (7.27); 18–45	China	Conservative	0.92	7	0.560
		usage time							0.170
	s-IAT-sex	frequency	167/156	25.31 (6.93); 18–45	China	Conservative	0.95	7	0.460
		usage time							0.220
Chen et al. (2020)	PPUS	frequency	782/0	21.32 (4.61); 18–48	China	Conservative	0.92	5	0.360
	PPUS	frequency	528/0	22.19 (5.17); 18–48	China	Conservative	0.95	6	0.470
Chen, Jiang et al. (2021)	PPCS	frequency	8845/0	25.82 (7.83); 18–65	China	Conservative	0.91	6.5	0.445
		usage time							0.305
	BPS	frequency							0.349
		usage time							0.227
	SCS	frequency							0.319
		usage time							0.218
Chen, Luo et al. (2021)	PPCS	frequency	4651/0	22.17 (4.33); 18–48	China	Conservative	0.93	6	0.390
	PPCS	frequency	695/0	25.39 (7.18); 18–48	China	Conservative	0.94	6	0.560
Chen, Wang et al. (2018)	PPUS	frequency	594/432	18.79 (0.93); 17–21	China	Conservative	-	5.5	0.570
Chen, Yang et al. (2018)	PPUS	frequency	466/342	18.54 (0.75); 17–22	China	Conservative	-	5.5	0.600
Degomez (2011)	ISST-Compulsivity	frequency	215/514	25; 18–66	USA	Permissive	-	5.5	0.199
Downing et al. (2014)	CIUS-adapted sexually explicit material	usage time	265/0	32.90 (12.50); 18-	USA	Permissive	0.19	6	0.380
Engel et al. (2019)	s-IAT-sex	usage time	630/564	42.64 (19.40); 18-	Germany	Permissive	0.83	5.5	0.454
Fernandez et al. (2017)	CPUI-9	frequency	76/0	22.27 (3.45); 18-	Malaysia	Conservative	0.90	6	0.320
Grubbs, Exline et al. (2015)	CPUI-9	usage time	228/103	19.50 (1.90); undergraduates	USA	Permissive	0.88	6	0.300
	CPUI-9	usage time	49/48	19.50 (1.30); undergraduates	USA	Permissive	0.89	6	0.320
	CPUI-9	usage time	136/72	31.80 (10.60); 18-	USA	Permissive	0.79	6.5	0.190
Grubbs et al. (2017)	CPUI-9	usage time	1021/498	19.30 (1.30); undergraduates	USA	Permissive	-	6	0.220
	CPUI-9	usage time	370/338 (5 not mentioned)	30.2 (9.9); 18-	USA	Permissive	-	6.5	0.180
Grubbs and Gola (2019)	CPUI-9 (compulsivity)	usage time	147/0	19.80 (3.70); undergraduates	USA	Permissive	-	5	0.320
	CPUI-9 (compulsivity)	frequency	297/0	46.50 (15.30); 18-	USA	Permissive	-	5.5	0.210
		usage time							0.190
	CPUI-9 (compulsivity)	usage time	433/0	33.5 (9.7); 18-	USA	Permissive	-	5.5	0.430
Grubbs, Kraus et al. (2019)	CPUI-9 (3 items)	frequency	862/599	44.80 (16.70); 18-	USA	Permissive	-	6.5	0.263
		usage time							0.144
Grubbs, Kraus et al. (2020)	CPUI-9 (compulsivity)	usage time	445/405	33.98 (9.87); -	USA	Permissive	-	6.5	0.441
Grubbs, Lee et al. (2020)	CPUI-9 (compulsivity)	frequency	946/478	43.92 (16.74); 18-	USA	Permissive	-	6.5	0.402
	BPS								0.335
Grubbs, Wilt, Exline, & Pargament et al. (2018)	CPUI-9	usage time	915/437	undergraduates;	USA	Permissive	0.90	5.5	0.180
	CPUI-9	usage time	387/406	18-	USA	Permissive	0.84	6.5	0.230
Harper and Hodgins (2016)	CPUI-9 (compulsivity)	frequency	86/105	21.05 (2.96); 17–38	Canada	Permissive	0.85	5.5	0.528
		usage time							0.302
	IP-CRIT	frequency							0.536
		usage time							0.389
Hatch et al. (2020)	PPUS	frequency	172/141 (4 not mentioned)	36.32 (10.20); -	USA	Permissive	-	6	0.460
		usage time							0.440
Klein et al. (2020)	s-IAT-sex	usage time	72/0	25.56 (4.45); 18– 40	Germany	Permissive	-	6	0.516
Kohut and Stulhofer (2018)	CPCS	frequency	314/0	second year of secondary school	Croatia	Conservative	-	5.5	0.400
	CPCS	frequency	197/0	second year of secondary school	Croatia	Conservative	-	5.5	0.240
Kor et al. (2014)	PPUS	frequency	834/886	39.52 (14.18); 18–70	Israel	Permissive	-	6	0.560
	PPUS	frequency	180/120	43.85 (14.34); 18–69	Israel	Permissive	-	6	0.610
Kraus et al. (2018)	SCS	frequency	265/0	21.56 (3.63); university students	USA	Permissive	0.82	6.5	0.300
Kraus et al. (2020)	BPS	frequency	698/360	32.94 (7.45); 18–55	USA	Permissive	0.82	6.5	0.391
		usage time							0.230
	BPS	frequency	338/132	44.90 (15.90); 18-	USA	Permissive	0.88	6.5	0.470
		usage time							0.330
	BPS	usage time	0/512	26.04 (6.07); 18–54	Poland	Conservative	-	6	0.070
	BPS	usage time	191/0	26.04 (6.07); 18–54	Poland	Conservative	-	6	0.170
Laier et al. (2014)	s-IAT-sex	usage time	0/51	21.83 (2.48); 18–29	Germany	Permissive	1.00	5.5	0.540
Laier et al. (2015)	s-IAT-sex	usage time	71/0	29.15 (6.22); 20–48	Germany	Permissive	0.00	6.5	0.185
Leonhardt et al. (2021)	CPUI-9 (compulsivity)	frequency	649/0	34.12 (10.46); 18–83	USA	Permissive	0.98	6.5	0.360
	CPUI-9 (compulsivity)	frequency	0/772	34.12 (10.46); 18–83	USA	Permissive	0.98	6.5	0.420
Leonhardt et al. (2018)	SCS	frequency	350/336	28.64 (7.85); 18–59	USA	Permissive	-	7	0.177
Levin et al. (2019)	CPUI-9	usage time	91/0	21.31 (3.99); 18–45	USA	Permissive	0.95	6.5	0.400
Lewczuk et al. (2020)	CPUI-9	frequency	485/395	43.69 (14.06); 18–69	Poland	Conservative	-	7	0.530
	BPS								0.440
Mennig et al. (2020)	OPDQ	usage time	537/156(7 not mentioned)	32.9 (12.6); 18-	Germany	Permissive	-	6.5	0.220
Pawlikowski et al. (2014)	s-IAT-sex	usage time	106/28	26.31 (8.31); -	Australia	Permissive	-	6	0.231
Rimington (2008)	SCS	usage time	134/123(5 not mentioned)	graduates and undergraduates; 18-	USA	Permissive	0.90	4.5	0.290
Rousseau et al. (2021)	CPCS	frequency	337/0	15.9 (0.52); high school sophomores	Croatia	Conservative	-	6	0.310
Sevcikova et al. (2021)	s-IAT-sex	frequency	129/20	55.44 (5.44); 50–77	Czech	Permissive	0.63	7	0.210
Shirk et al. (2021)	PPUS	frequency	172/0	33.9 (8.53); -	USA	Permissive	-	6.5	0.390
Snagowski and Brand (2015)	s-IAT-sex-craving	frequency	123/0	23.79 (5.10); 18-	Germany	Permissive	1.00	5.5	0.227
		usage time							0.198
Stark et al. (2017)	s-IAT-sex	usage time	48/0	25.96 (4.96); 18–44	Germany	Permissive	1.00	6	0.330
	s-IAT-sex	usage time	0/47	24.94 (5.09); 18–42	Germany	Permissive	1.00	6	0.330
Volk et al. (2019)	CPUI-9	frequency	90/0	32.1 (7.79); 18–49	USA	Permissive	-	7	0.242
	CPUI-9	frequency	0/89	32.1 (7.79); 18–49	USA	Permissive	-	7	0.026
Volk et al. (2016)	CPUI-9	usage time	187/171	34.13 (10.61); 18–66	USA	Permissive	0.88	7	0.200
Wéry et al. (2016)	s-IAT-sex	frequency	388/0	28.74 (8.80); 18–72	Belgium	Permissive	0.89	6.5	0.235
Wetterneck et al. (2012)	SCS	usage time	152/342	29.50 (9.00); 18–66	USA	Permissive	0.90	5.5	0.380
Wilt et al. (2016)	CPUI-9	usage time	732/338	19.33 (2.53); undergraduates	USA	Permissive	-	6	0.220

Note. ^a The higher the total score, the higher the quality of the research. PPU = Problematic Pornography Use; QPU = Quantity of Pornography Use; s-IAT-sex = Short Internet Addiction Test Adapted to Online Sexual Activities; ISST- Compulsivity = the compulsivity factor of Internet Sex Screening Test; IP-CRIT = adapted DSM-5 Internet pornography addiction criteria; BPS = Brief Pornography Screen; CIUS-adapted sexually explicit material = Compulsive Internet Use Scale adapted sexually explicit material; CPCS = Compulsive Pornography Consumption Scale; CPUI-9 = Cyber Pornography Use Inventory-9; PPCS = Problematic Pornography Consumption Scale; PPUS = Problematic Pornography Use Scale; SCS = Sexual Compulsivity Scales; OPDQ = Online Pornography Disorder Questionnaire.

Data Analysis

We calculated r coefficients for effect-size estimates using “metafor” (version 2.4– 0) and “meta” (version 4.12– 0) for R software (version 4.0.2). Effect sizes of at least 0.50 were considered large, 0.30 were considered moderate, and 0.10 were considered small effects (Cohen, 1992). Effect sizes were computed using random-effects models. The Q statistic and I² were used to test heterogeneity of effect sizes. Subgroup analyses with categorical variables (e.g., sexual cultural context, gender, sexual orientation, measurements) and meta-regression analyses with continuous variables (e.g., quality of study, participants’ age, published year, and sample size) were performed to explain observed heterogeneity in effect sizes. In order to evaluate the influence of each study on the overall effect size, sensitivity analyses were conducted using the leave-one-out method and the influential case diagnostics method [i.e., DFFITS value, Cook’s distances, covariance ratios, estimates of τ² and test statistics for (residual) heterogeneity] (Alemayehu & Hailemariam, 2020; Gallo et al., 2020; Polanczyk et al., 2014; Raboisson et al., 2020; Da Silva Chaves et al., 2018; Simental-Mendía et al., 2018; Wolfgang, 2010). To test publication bias, we used funnel plots, Egger’s regression test, and Rosenthal’s Fail-safe N to assess symmetry.

Results

Overall Estimation of Effect Sizes between QPU and PPU

The current meta-analysis included 61 studies with 74,880 participants and 82 independent effect sizes as 15 studies consisted of multiple samples. The effect size of the overall association was r = 0.34 (95%CI = 0.31, 0.37) with the random-effects model. The Q statistic was significant (Q₍₈₁₎ = 1171.73, p < .001) and I² value was 94.52%, suggesting that there was significant heterogeneity among effect size estimates for the studies in this meta-analysis. In the forest plot of the leave-one-out sensitivity analysis for random-effects models (see Supplementary Material 1, Figure S1), we can see how the overall effect estimates change with one study omitted each time. All analyses were robust in the leave-one-out sensitivity analysis because all of the recalculated correlation coefficients remained around 0.34 (95%CI = 0.32, 0.35). Again, the plot of the influence analysis (see Supplementary Material 1, Figure S2) indicated that it was not necessary to remove any data, since no primary study reached the cutoffs proposed by Viechtbauer and Cheung (Harrer et al., 2021; Viechtbauer & Cheung, 2010; Wolfgang, 2010).

Publication Bias

To test for publication bias in the examined manuscripts, a funnel plot was generated (Figure 2). Most plots were at the top and relatively evenly distributed near the principal axis. However, the funnel plot appeared a little asymmetrical at the top, and the trim-and-fill analysis identified 12 putative missing studies on the right side of the distribution (Figure 3), suggesting a possible publication bias. Considering the subjective interpretation of the funnel plot, we also used Egger’s regression test and Rosenthal’s Fail-safe N to assess for symmetry. Based on the result of the Egger’s test (Z = −1.15, p = .248), no publication bias was detected. As for Fail-safe N (p < .001), the result showed that at least 202,807 studies with the opposite conclusions would be required to overturn the findings of this meta-analysis. In sum, multiple inspection methods provide little evidence for publication bias in this meta-analysis.

Figure 2. The original funnel plot of the effect sizes (K = 82).

Figure 3. The adjusted funnel plot using the trim-and-fill method (K = 82).

Subgroup Analysis

Subgroup analyses was conducted to examine characteristics that might explain variance in effect sizes across studies. Results of subgroup analyses are displayed in Table 2. As expected, there were significant differences in between-group correlations (Q_B = 12.66, p = .002). The scales in the framework of addiction showed the strongest associations with QPU (r = 0.39). Moderation analyses showed that the effect size of frequency was significantly larger than that for usage time (Q_B = 14.49, p < .001), and multi-item measurements had stronger relationships than single-item measurements (Q_B = 15.44, p < .001).

Table 2. Summary of the results of the sub-group meta-analysis.

Moderator	Subgroup	k	Effect size (95% confidence) interval)			Q	I^2 (%)	p
			effect size(r)	lower	upper
Measurement of PPU	Developed based on the addiction framework	45	0.388	0.358	0.417	547.57	92.0	0.002
	Developed based on the obsessive-compulsive framework	22	0.323	0.285	0.361	144.17	85.4
	Developed based on the self-perceived PPU framework	20	0.280	0.214	0.344	284.47	93.3
Measurement of QPU	Frequency	51	0.381	0.347	0.415	967.9	94.8	< 0.001
	Use time	50	0.293	0.263	0.323	502.94	90.3
Measurement method	Single-item	78	0.312	0.280	0.344	1728.73	95.5	< 0.001
	Multi-item	23	0.426	0.379	0.470	320.42	93.1
Sexual orientation^a	Mostly sexual minority	2	0.309	0.116	0.479	2.45	59.1	0.718
	Mostly heterosexual	46	0.343	0.314	0.371	418.30	89.2
Sample type	Treatment-seeking	5	0.346	0.295	0.395	29.31	86.4	0.916
	Community	77	0.343	0.313	0.373	1126.88	93.3
Gender	Men	38	0.376	0.340	0.412	361.89	89.8	0.506
	Women	13	0.408	0.321	0.488	122.30	90.2
Sexual culture	Conservative	15	0.413	0.348	0.474	313.18	95.8	0.021
	Permissive	67	0.329	0.298	0.358	844.88	92.2
Sexual culture × Gender	Conservative × Men	13	0.426	0.355	0.493	-	-	0.246
	Conservative ×Women	7	0.425	0.324	0.517	-	-
	Permissive ×Men	25	0.342	0.284	0.398	-	-
	Permissive ×Women	6	0.383	0.262	0.493	-	-
Sexual culture × Measurement of PPU	Conservative × Addiction framework	11	0.444	0.380	0.504	-	-	< 0.001
	Conservative × Obsessive-compulsive framework	4	0.251	0.126	0.369	-	-
	Conservative × Self-perceived PPU framework	2	0.459	0.285	0.604	-	-
	Permissive × Addiction framework	34	0.364	0.322	0.405	-	-
	Permissive × Obsessive-compulsive framework	18	0.338	0.282	0.392	-	-
	Permissive × Self-perceived PPU framework	18	0.262	0.202	0.320	-	-

Note. ^a The ratio of heterosexual-sexual minority participants was reported in Table 1; if the ratio was greater than 0.5, it was considered as a mostly heterosexual sample. If not, it was considered as a mostly sexual minority sample.

As can be seen in Table 2, the overall magnitudes of the correlation coefficient did not significantly vary across gender (female vs male; Q_B = 0.44, p = .506), sexual orientation (mostly sexual minority vs mostly heterosexual samples; Q_B = 0.13, p = .718) and type of sample (community vs treatment-seeking/clinical; Q_B = 0.01, p = .916).

Results suggested that cultural context significantly contributed to the heterogeneity in the associations (Q_B = 5.33, p = .021). The correlation between QPU and PPU was stronger in more conservative countries (r_conservative = 0.41 vs r_permissive = 0.33). Although no significant interaction was observed between sexual culture context and gender (p = .246), there was a significant interaction of sexual culture context and theoretical frameworks of PPU (p < .001). In conservative sexual contexts, QPU showed robust association with self-perceived PPU, and weakest relations with PPU based on the obsessive-compulsive framework. The outcomes of meta-regression analyses showed that the magnitude of the r coefficient did not vary according to study quality, participant age, first age exposed to pornography, publication year, and sample size (all ps > 0.076).

Discussion

Given previously reported variability in associations between QPU and PPU (Bőthe, Tóth-Király et al., 2018; Chen, Ding et al., 2018; Fernandez et al., 2017; Gola et al., 2016; Grubbs, Exline et al., 2015; Grubbs et al., 2017; Grubbs & Gola, 2019; Grubbs, Kraus et al., 2019; Grubbs, Wilt, Exline, Pargament et al., 2018; Volk et al., 2016; Wéry et al., 2016), the present meta-analysis examined the associations between QPU and PPU considering the roles of potential moderators. There was an overall significant, positive and moderate relationship between QPU and PPU (r = 0.34, k = 82). Moreover, the association of QPU and self-perceived PPU was close to that reported by Grubbs and colleagues (r = 0.28, k = 20 vs r = 0.27, k = 10, respectively) (Grubbs, Perry et al., 2019). The relationship appeared to be moderated by other factors leading to differences in strengths of the relationships. Subgroup analyses showed that measurements of PPU in different conceptual frameworks may contribute importantly to the strength of the relationships. Frequency of use seemed more robustly related to PPU than usage time. Additionally, QPU and PPU were more strongly associated in more conservative countries in comparison to more sexually permissive countries, regardless of participants’ gender. Associations between QPU and PPU did not vary by gender, treatment-seeking status, sexual orientation, age, publication year, or sample size.

Measurements and Theoretical Frameworks of PPU

When examining associations between QPU and PPU using different measures assessing PPU, effect sizes obtained when using the scales based on addiction frameworks (i.e., PPUS, PPCS) were the largest. These larger effect sizes may derive from the fact that these scales include essential addiction factors, such as salience or tolerance, that may be more strongly related to QPU than other aspects of PPU included in other conceptualizations of PPU (e.g., tolerance refers to the process whereby increased amounts of the activity are required to achieve the former mood modifying effects). Scales developed to assess self-perceived addiction (e.g., CPUI-9) appear to be somewhat less correlated with QPU. Considering that excessive QPU can be considered as an objective indicator of dysregulated PPU (Cooper et al., 1999), it was reasonable to find that there were more robust relationships between objective behavior frameworks (i.e., the addiction framework and the obsessive-compulsive framework) and QPU, than frameworks focusing on perceptions of PPU (i.e., self-perceived addiction framework), since perceived addiction refers to subjective distress of inability to control one’s own consumption, regardless of the quantity of actual pornography use (Grubbs, Volk et al., 2015). The lesser focus on addictive features and greater focus on subjective addiction may contribute to the weaker associations between QPU and PPU, although some studies using self-perceived PPU scales found particularly strong relationships with frequency of use (Lewczuk et al., 2020).

Prior evidence also supported the smallest effect size in the self-perceived addiction framework. Longitudinal findings suggest that participants’ self-perceived addiction may not predict frequency of pornography use over time (Grubbs, Wilt, Exline, & Pargament 2018), and low frequency of use may also result in conflict and self-reported problems in pornography use (Chen, Jiang et al., 2021). However, individuals with PPU reported both high levels of QPU and features of PPU including and tolerance, and relapse (Bőthe, Lonza et al., 2020). Therefore, a practical implication for PPU, assessment would be that, taking into account the possibility of specific individuals overpathologizing their own behavior, items focusing on objective behavior patterns (e.g., “I have had many unsuccessful attempts to reduce my pornography use”) would likely have greater sensitivity compared to items focusing on beliefs or perception (e.g., “I believe I am addicted to pornography”).

Frequency and Usage Time of QPU (Single or Multi-item)

In line with the findings of a recent study (Lewczuk et al., 2020), the current study revealed that frequency of use was a more robust quantitative indicator of PPU than usage time. As Busby et al. proposed that pornography is a multidimensional construct (Busby et al., 2017, 2020), the assessment of frequency should be more diverse and systematic (Bőthe, Tóth-Király et al., 2018; Chen, Yang et al., 2018; Rimington, 2008; Wéry et al., 2016). Since participants may have different standards as to what constitutes pornography (Kohut et al., 2020; McKee et al., 2020), when pornography research uses simple single-item measures (e.g., How often have you used pornography during the last 6 months?), some participants may refer to very mild types of sexual media, whereas others will be thinking about extremely graphic types of sexual media (Busby et al., 2020). Furthermore, results based on single-item measures may underestimate the prevalence of such experiences and accuracy may be biased by stereotypes about behavioral and sexual experiences (Shaughnessy & Byers, 2013). Therefore, researchers are encouraged to use multi-item or multi-dimensional measures of QPU that consider frequency, usage time, content, and other features [e.g., motivations (Bőthe, Tóth-Király, Bella et al., 2021; Grubbs, Wright et al., 2019)]. In order to assess pornography use in a more comprehensive manner, measures with demonstrated content validity and a definition of pornography should be used (Kohut et al., 2020), to provide more consistency between different studies (Grubbs et al., 2020).

Conservative and Permissive Sexual Cultural Contexts: Interaction with Theoretical Frameworks of PPU

Concerning sexual cultural contexts, there was a significantly larger effect in conservative than in sexually permissive countries in the associations between QPU and PPU. In the conservative sexual contexts, QPU had the strongest association with self-perceived PPU. However, in the sexually more permissive countries, this relationship was relatively stable without significant variations in different PPU theoretical frameworks. In the current study, conservative/permissive sexual cultural contexts were classified based on acceptance of homosexuality, abortion and divorce scores from the WVS. As Young proposed (2001, 2008), individuals’ pornography use in more conservative cultures may progress more rapidly to the feeling of PPU (i.e., self-perception of PPU), perhaps due to specific factors (e.g., cultural shame, not knowing how/where to receive help) that may limit behavioral changes at earlier time points. On the other hand, in conservative contexts, individuals may feel that pornography use is in opposition to their cultural values (Grubbs & Perry, 2019); thus, any pornography use may be considered problematic. Cultural context and other factors (cultural aspects relating to religion) warrants further research, as do jurisdictional considerations that may be more independent of culture (e.g., with respect to regulation of the internet).

Gender Differences: Interaction with Conservative/permissive Cultures

Previous studies have suggested that relationships between QPU and PPU are weaker among men than women, especially in treatment-seeking settings (Gola et al., 2016; Lewczuk et al., 2017). However, in the current meta-analysis, gender did not significantly moderate the magnitude of the association between QPU and PPU. Although the prevalence of PPU in women appears to be lower than that in men, a considerable number of women experience PPU (Baranowski et al., 2019; Green et al., 2012). From the addiction literature (Lev-Ran et al., 2013), we know that women usually demonstrate lower prevalence of addictive behaviors than men, but do show a similar symptoms. Several studies have demonstrated that PPU scales have a similar factor structure in men and women, but men show higher levels of PPU in general (Bőthe, Tóth-Király, Demetrovics et al., 2021; Bőthe, Tóth-Király et al., 2018; Chen, Wang et al., 2018), suggesting that men and women may have similar features of PPU. In sexually conservative cultures, the association between QPU and PPU was higher among both men and women. In more conservative cultures, non-marital or recreational sexual activities are often stigmatized, and members’ sexuality are controlled (Guo, 2019), which makes it riskier for these individuals to use pornography and may exacerbate their conflict.

Treatment-seeking Status

Although no significant differences were found in effect sizes in treatment-seeking and community samples, recent research suggested that withdrawal and tolerance may be particularly important factors in treatment-seeking individuals (Chen, Luo et al., 2021), and these features may underlie frequent pornography use (Sussman & Sussman, 2011). That is, QPU may be more strongly associated with PPU among individuals who sought treatment for compulsive use, and this may reflect processes of tolerance and withdrawal. Primarily non-clinical samples have been used in most studies of PPU; among the 61 included studies, only 5 studies covered treatment/help-seeking participants, which may explain the non-significant difference between the treatment-seeking and community samples. More research is needed to investigate relationships between QPU and PPU in clinical/ treatment-seeking populations, as well as to assess the relevance of tolerance and withdrawal for behavioral addictions more broadly.

Sexual Orientation

Sexual orientation may potentially influence relationships between QPU and PPU (Brand, Brand, Antons et al., 2019). Sexual minority groups have been reported to engage in more frequent pornography use, and spend more time with it (Downing et al., 2017), possibly because they have more obstacles to find sexual or romantic partners (Bőthe et al., 2019). Nevertheless, no significant moderation effects of sexual orientation on associations between QPU and PPU were observed, which may be due to the fact that sexual excitability and pleasure acquired from pornography use can be similar for sexual minority and heterosexual individuals. For example, Laier et al. found that sexual excitability from pornography use seemed to be independent from gender and sexual orientation (Laier et al., 2014, 2015). Brain imaging findings comparing neural activations of homosexual and heterosexual men when presenting sexually relevant cues also supported this notion (Paul et al., 2008).

Limitations and Future Studies

The present study was not without limitations. First, the studies included in the meta-analysis mostly used self-report scales, and no clinical examinations were conducted, which may bias the results. Second, available studies largely lacked sizable numbers of sexual minority individuals or did not state explicitly participants’ sexual orientations. Thus, there may have been insufficient sample sizes to evaluate main and interaction effects, including interactions between gender and sexual orientation, or sexual cultural context and sexual orientation. Third, although some studies suggested links between religiosity and self-perceived PPU (Bradley et al., 2016; Grubbs, Exline et al., 2015; Grubbs & Perry, 2019; Wéry et al., 2016), few studies explicitly compared atheist and religious samples, which limits possibilities for more detailed analyses. It may be important for future studies to conduct more investigations in sexual minority samples and religious samples in order to obtain enough study findings to carry out future meta-analysis to generate a more systematic understanding of interactions between these variables.

Additionally, studies typically did not examine pornography content, and some pornographic content [e.g., objectification of and violence toward women (Bridges et al., 2010) and themes of rape, incest and racism (Fritz et al., 2021; Rothman et al., 2015)] may conflict with moral values in ways that may be independent of religion or sexual orientation. Thus, future studies should consider pornography content. Lastly, although the CPUI-9 and CPUI-3 mainly assess self-perceived PPU, they also include items related to compulsive use; in addition, there might be overlap between the addiction and obsessive-compulsive conceptualizations. Therefore, it was not possible to examine these different frameworks completely separately. Although the scales were developed using different theoretical frameworks, they are strongly related; for example, the BPS and PPCS BOTH assess PPU but from a different perspective and were strongly related in previous work (Chen, Jiang et al., 2021).

Future studies should consider religion and morality in a more detailed manner to understand what moral incongruence may mean and how it may relate to different sexual cultural contexts. It will be important in future studies to examine relationships between QPU and PPU in longitudinal studies, considering further moderating and mediating factors. For instance, viewing sexually explicit materials may result in more permissive sexual attitudes (Bishop, 2015; Peter & Valkenburg, 2016), and conservative beliefs about sexual behaviors seem to be particularly related to self-perceptions of PPU for many people. Thus, individuals’ attitudes toward sexuality may be both a mediator and a moderator that should be investigated. A consolidation of theoretical understandings and assessments of pornography use and PPU would be a necessary pre-requisite for future research in this domain. We believe these are key considerations that should underlie current and future efforts to develop or retain measures that are useful and accurate in the assessment of QPU and PPU and based on solid theoretical frameworks and strong empirical evidence. Direct measures of cultural characteristics should also be considered in future studies. Additionally, individual differences such as personality characteristics (e.g., sexual sensation seeking) and level of self-esteem may also be relevant moderators between QPU and PPU (Borgogna, McDermott, et al., 2020; Chen, Yang et al., 2018). However, given the lack of sufficient number of studies, we were not able to conduct subgroup analyses with these variables.

Conclusion

In general, QPU had a positive, moderate association with PPU. However, the strength of this association was dependent on several factors (i.e., moderators), such as the measurements of PPU or the sexual cultural context. The magnitude of the association was higher in the case of scales using the addiction framework, and for the more sexually conservative countries. Contrary to expectations, gender, sexual orientation, and treatment-seeking status did not moderate the associations between QPU and PPU.

Author Contributions

Conceptualization, Lijun Chen, Huijuan Wu; methodology, Lijun Chen, Xiaoliu Jiang, Beáta Bőthe; formal analysis, Lijun Chen, Xiaoliu Jiang, Qiqi Wang; data collection, Xiaoliu Jiang, Qiqi Wang; resources, Marc. N. Potenza, Lijun Chen; data curation, Lijun Chen, Beáta Bőthe; writing—original draft preparation, Xiaoliu Jiang; writing—review and editing, Lijun Chen, Xiaoliu Jiang, Beáta Bőthe, Marc. N. Potenza, Huijuan Wu; visualization, Xiaoliu Jiang, Qiqi Wang; supervision, Huijuan Wu, Marc. N. Potenza; project administration, Lijun Chen, Huijuan Wu.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplemental data for this article can be accessed on the publisher’s website.

Additional information

Funding

The research was supported by the National Social Science Foundation of China (Grant No. 19BSH117). BB was funded by a postdoctoral fellowship award by Team SCOUP – Sexuality and Couples – Fonds de recherche du Québec, Société et Culture. MNP’s involvement was supported by the Connecticut Council on Problem Gambling. 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.

Notes

¹ In order to reduce the collinearity deriving from the PPU measurements that included items on both QPU and PPU, we excluded PPU scales that included questions about the frequency of use and usage time. For example, the Internet Sex Screening Test (ISST) comprised items on QPU; thus, we only included the papers in the present study that reported the compulsivity dimension of the ISST separately (not the total score of the ISST).

² For example, the Sexual Addiction Screening Test-Revised (SAST-R) comprised items that assessed sexual activities in general. Therefore, we did not include studies using the SAST-R.

³ http://www.worldvaluessurvey.org/wvs.jsp

⁴ The rating criterion was that studies reported associations between QPU and PPU (total score or at least one dimension of PPU).

⁵ The rating criterion was the breadth of sample distribution. For instance, given that the sample distribution for online recruitment is much broader than for course recruitment, we rated online survey 1, and other convenient sampling such as course recruited as 0.5.

⁶ The scoring criterion was whether the study reported the actual Cronbach’s alpha.