Understanding Binge-Watching: The Role of Dark Triad Traits, Sociodemographic Factors, and Series Preferences

Abstract

Binge-watching has become a dominant mode of media consumption, yet its psychological underpinnings remain insufficiently explored. This study investigates the influence of dark triad personality traits, sociodemographic factors, and TV series preferences on binge-watching behavior. A Portuguese version of the Binge-Watching Engagement and Symptoms Questionnaire (BWESQ) was validated using a confirmatory factor analysis (CFA), demonstrating strong psychometric properties. Measurement invariance was confirmed across genders through a multigroup CFA, testing configural, metric, scalar, and error variance levels. Model reliability, convergent validity, and discriminant validity were assessed using Cronbach’s alpha, composite reliability, and the average variance extracted (AVE). Multiple linear regression analyses identified significant predictors of binge-watching, including gender, age, professional status, TV series preferences, Machiavellianism, and psychopathy. These findings underscore the role of personality traits and demographic factors in shaping binge-watching behavior, offering insights that may inform psychological interventions to prevent excessive or problematic viewing patterns.

1. Introduction

Media consumption has shifted markedly from traditional broadcast schedules to flexible, on-demand streaming platforms [1]. Unlike traditional TV, where content follows fixed programming, online services allow users to choose what and when to watch, leading to new viewing behaviors [2]. With accessible interfaces, affordable plans, and vast libraries available on most devices, streaming platforms have become deeply embedded in everyday life [3].

Streaming technology has shifted viewing patterns from weekly episode releases to full-season availability [3,4], driving the rise of binge-watching, especially among young adults [3,5,6]. Defined as watching multiple episodes in one sitting [4], binge-watching can be hard to distinguish from casual or marathon viewing [7], with studies typically defining it by episode count, frequency, and content type [5,8]. This behavior is complex, involving both gratification and compensation [7]. It can manifest as (1) a rewarding experience fulfilling needs and desires, or (2) problematic behavior, linked to negative consequences and risk factors like age, maladaptive coping, impulsivity, automatic behavior, and mental health issues [3,9].

1.1. Literature Review

Variables Associated with Binge-Watching

Social factors strongly motivate binge-watching, as people watch multiple episodes to build connections, join discussions, and gain peer acceptance [5,10]. Shim and Kim [11] show that recommendations from others increase the motivation to watch. Panda and Pandey [2] identify social engagement, escapism, advertising, and accessibility as key drivers. Interestingly, negative gratification, like anxiety after binge-watching, can lead to more frequent behavior and increased dependency [2]. Ahmed [12] notes that foreign-dubbed dramas, action films, comedies, and documentaries are most popular.

Research links being female with higher binge-watching engagement, especially in session frequency, intensity, and the loss of control [9,13,14]. Conversely, men often have longer binge-watching sessions [12,13,14,15,16,17]. Millennials, born between 1980 and 2000, are the main streaming platform subscribers and binge-watch more than other age groups [5,10].

Research shows binge-watching can be enjoyable [18], but also resembles behavioral addictions, like internet or smartphone overuse [8]. Excessive binge-watching shares addiction traits, such as difficulty controlling viewing time [10,19]. The convenience of streaming platforms can promote overuse, with individual differences and risk factors determining whether it becomes problematic [19,20].

To assess whether non-chemical addictions are genuine or metaphorical, researchers compare them to clinical criteria for substance addictions [21,22]. This approach has identified behavioral addictions like “television addiction” and “amusement machine obsession”. Griffiths [22] states that all addictions share key components: salience, mood modification, tolerance, withdrawal symptoms, conflict, and relapse. Kilian et al. [23] explored its behavioral and neural correlates, suggesting that, like substance use disorders, excessive binge-watching may involve deficits in cognitive control, though this remains underexplored [23].

1.2. Binge-Watching and Dark Triad

Binge-watching is an automatic behavior in which viewers continue watching without conscious intent [24]. As defined by Costa and McCrae [25], personality encompasses individual differences in emotions, interests, attitudes, motivations, and interpersonal or experiential styles. Numerous studies have linked personality traits to preferences across various contexts, including media consumption activities, such as reading books or attending concerts [26]. These traits also influence preferences for specific genres and categories in entertainment, such as music, television programs, films, books, magazines, and museums [27,28].

Personality traits influence internet usage patterns, shaped by individual needs and internet services [29,30]. People who are high in dark triad traits focus on short-term activities, like entertainment and shopping, avoiding tasks requiring sustained effort or long-term interests, such as online learning or detailed information searches [5,29,31]. Narcissism, marked by a sense of superiority and entitlement, correlates with higher engagement in problematic social media, online gaming, and internet use [32,33,34].

Machiavellianism, characterized by manipulativeness, links to problematic social network use and gaming, as it often serves as a coping mechanism for negative emotions [35,36]. Psychopathy, defined by impulsivity and a lack of empathy, is also associated with problematic internet use as a maladaptive coping strategy [32,37].

Subclinical personality traits often indicate a predisposition to behavioral addictions [38]. Narcissism and psychopathy are linked to both substance and non-substance addictions, but Machiavellianism shows no such association [39]. Among narcissistic traits, “feeling privileged” predicts internet addiction tendencies, while “superiority” correlates with social isolation [40]. Binge-watchers show higher neuroticism, lower agreeableness, reduced conscientiousness, and less openness [5], with compulsive individuals more likely to experience negative emotions, low stress tolerance, and low self-esteem [15,41].

Despite the widespread popularity of binge-watching, research on its psychological correlates remains limited, as most studies focus primarily on the motivations behind this behavior [10]. Moreover, there is insufficient research exploring the personality traits associated with binge-watching. Understanding how personality traits correlate with binge-watching helps identify at-risk individuals (e.g., impulsive or emotionally detached viewers) and tailor interventions. Traits like psychopathy or neuroticism reveal motivations—escapism, thrill-seeking, or poor self-regulation—that drive compulsive viewing. This knowledge allows mental health professionals to address root causes (e.g., stress, loneliness) and design targeted strategies, such as mindfulness for anxiety-driven viewers. For streaming platforms, insights into personality-driven habits can inform ethical design, like prioritizing “protective” genres (documentaries) or adding break reminders during intense content. Public health campaigns can educate high-risk groups about genre-specific risks, while reducing stigma by framing binge-watching as a behavior shaped by measurable traits, not moral failure. Ultimately, linking personality to media use balances healthier engagement with preserving entertainment’s value, advancing both psychological research and practical solutions.

While recent studies examine its relationship with the Big Five personality traits [25], no research to date has explored the role of the dark triad traits or preferences for specific series genres in explaining binge-watching [5]. This study seeks to address this gap by analyzing how dark triad personality traits and series preferences contribute to binge-watching behavior.

2. Materials and Methods

2.1. Study Design

This study employed a cross-sectional quantitative design to examine the relationships between variables. Data were collected at a single point in time using standardized self-report questionnaires. The quantitative approach allowed for the measurement and statistical analysis of associations between variables, providing a snapshot of patterns and potential correlations within the target population.

2.2. Ethics

This study adheres to the ethical principles outlined in the Declaration of Helsinki [42]. Ethical approval for the research was granted by the Scientific Council of the Universidade Católica Portuguesa, Braga, Portugal. Additionally, permission was obtained from the original authors for the use of the instruments in the study.

All the data collected were stored securely using Google Forms, which ensures data protection through encryption and restricted access. The participants’ privacy and confidentiality were maintained throughout the study, and only authorized personnel had access to the data. The research design and data handling procedures were in full compliance with relevant ethical standards to protect participants’ rights and ensure the integrity of the research process.

2.3. Study Population

The sample consists of individuals from the general population, representing a diverse group in terms of gender, age, marital status, and employment. It includes participants from various life stages, including those who are single, married, cohabiting, divorced, or widowed. Additionally, the sample encompasses individuals with different employment statuses, such as those who are actively employed, students, retired, or unemployed. This broad representation of the general population ensures a varied and comprehensive perspective.

2.4. Selection and Recruitment

The participants met the following inclusion criteria: Portuguese nationality, aged 18 or older, and regular TV series viewing. All the participants received detailed information about the study’s objectives and assurances of anonymity and confidentiality and completed an informed consent form.

2.5. Data Collection

The data collection employed a non-probabilistic snowball sampling method. The data collection took place between July and October 2024.

2.5.1. Procedures

The Binge-Watching Engagement and Symptoms Questionnaire (BWESQ) was translated and adapted following the International Test Commission (ITC) guidelines [43] and the back-translation method [44]. Two bilingual translators, one a psychologist and the other from the social sciences, independently translated the instrument from English to Portuguese. A third bilingual translator, also a psychologist, reconciled the translations, and the back-translation was reviewed against the original English version by the first translator to ensure linguistic and cultural equivalence. No discrepancies were identified between the back-translated and original versions. The Portuguese version of the scale is available in Appendix D, Appendix E and Appendix F.

2.5.2. Instruments

The questionnaire used in this study is detailed in Appendix A, Appendix B and Appendix C, in English and Appendix D, Appendix E and Appendix F in Portuguese.

The Sociodemographic Questionnaire

The sociodemographic questionnaire includes four questions: gender (0—male; 1—female), age (18–28 (0); 29–39 (1); 40–50 (2); 50–60 (3); and >60 (4)), professional status (student—0; active worker—1; unemployed—2; retired—3), and marital status (single, no dating relationship—0; single, but in a dating relationship—1; married or in a de facto relationship—2; divorced/separated—3; widowed—4).

The Questionnaire About Television Series Preferences

The questionnaire includes a question designed to assess the participants’ general preferences for the TV series they watch. Response options are categorized as either (1) calming (calm and relaxing) or (2) stimulating (active and/or violent). Additionally, the participants are asked to specify the genres they watch most frequently, with options including action, drama, horror, comedy, science fiction and fantasy, romance, and documentaries. In this study, calming series are defined as those that are primarily calming, slow-paced, and emotionally soothing. These often include gentle narratives with minimal conflict, and are designed to help viewers relax (e.g., feel-good dramas, cozy mysteries, or nature documentaries with a peaceful tone). Stimulating series refer to those that are fast-paced, high in suspense or action, and emotionally arousing—including series with violence, intense conflict, or adrenaline-inducing scenarios (e.g., thrillers, action-packed dramas, or crime series).

The Binge-Watching Engagement and Symptoms Questionnaire (BWESQ)

The Binge-Watching Engagement and Symptoms Questionnaire (BWESQ; Flayelle et al. [45]) consists of 40 items, which are organized into seven subscales: loss of control (items 11, 12, 15, 23, 29, 32, and 36; α = 0.83); engagement (items 1, 9, 18, 26, 27, 30, 35, and 39; α = 0.79); dependency (items 6, 10, 16, 25, and 31; α = 0.77); desire/savoring (items 2, 3, 4, 5, 7, and 21; α = 0.75); positive emotions (items 8, 24, 28, 33, and 38; α = 0.62); binge-watching (items 14, 17, 19, 20, 22, and 34; α = 0.79); and pleasure preservation (items 13, 37, and 40; α = 0.63). The questionnaire is rated on a 4-point Likert scale, where 1 = Strongly Disagree and 4 = Strongly Agree. A higher total score indicates a more compulsive behavior toward watching TV series.

The Dark Triad Dirty Dozen

The Dark Triad Dirty Dozen (DTDD; Jonason & Webster [46]) is a self-report instrument designed to assess the dark triad of personality traits. It consists of 12 items, organized into three subscales: Machiavellianism (items 1, 2, 3, and 4); psychopathy (items 5, 6, 7, and 8); and narcissism (items 9, 10, 11, and 12). The original version of the scale demonstrated strong psychometric properties, with Cronbach’s alpha values of α = 0.79 for Machiavellianism, α = 0.77 for psychopathy, α = 0.77 for narcissism, and α = 0.86 for the overall scale. The Portuguese version of the instrument [47] maintains the same structure, with 12 items divided into the three subscales: Machiavellianism, characterized by manipulation and a lack of internalized morality; psychopathy, marked by emotional insensitivity and a lack of remorse; and narcissism, associated with selfishness and a grandiose self-image. The scale is rated on a 5-point Likert scale, ranging from 1 = Strongly Disagree to 5 = Strongly Agree, with higher scores indicating greater psychopathological traits. The Portuguese version demonstrated satisfactory psychometric properties for the Machiavellianism (α = 0.58) and psychopathy (α = 0.54) subscales, and strong psychometric properties for the narcissism subscale (α = 0.74).

2.6. Data Analyses

Descriptive analyses are performed to examine sociodemographic characteristics and television series preferences. The normality of the BWESQ items is assessed using skewness (SI < 3) and kurtosis (KI < 10) indexes, indicating no severe violations of normality [48]. Multicollinearity is evaluated through tolerance (>0.100) and the variance inflation factor (VIF) (<10) [49].

Confirmatory factor analyses (CFAs) are conducted for the Portuguese sample to evaluate the fit of the seven-factor model derived from the original BWESQ validation [45]. The model’s fit is assessed using the root mean square error of approximation (RMSEA), the comparative fit index (CFI), the incremental fit index (IFI), and the standardized root mean square residual (SRMR). An excellent model fit is indicated by CFI and IFI values of ≥0.95, an RMSEA of ≤0.05, and an SRMR of ≤0.05 [50]. Acceptable fit is defined as CFI and IFI values of ≥0.90, an RMSEA of ≤0.08, and an SRMR of ≤0.10 [51]. The Satorra–Bentler chi-square (X²), general model significance (p), and relative chi-square (X²/df) are reported; however, X is interpreted with caution due to its sensitivity to sample size [52].

To assess the factor structure of the BWESQ across genders, multigroup CFAs are conducted, testing four levels of measurement invariance: configural (whether items load on the same factor across groups), metric (whether item factorial loadings are equal across groups), scalar (whether item intercepts are equal across groups), and error variance (whether item measurement errors are equal across groups). The progressive constrained models are assessed by comparing nested models (Δ) using RMSEA, CFI, and SRMR changes. A change of ≥0.01 in CFI, ≥0.015 in the RMSEA, or ≥0.03 in the SRMR indicates a significant decrease in model fit when testing for measurement invariance [53].

Pearson correlations are calculated for continuous variables, and Spearman correlations are used when one or more variables are ordinal or nominal. Correlations between 0 and 0.3 are considered weak, 0.3 to 0.5 moderate, 0.5 to 0.7 strong, and 0.7 to 1 very strong, whether positive or negative [54].

To evaluate the model’s reliability, convergent and discriminant validity are assessed using Cronbach’s alpha coefficients, composite reliability (CR, with values ≥0.70 suggesting adequate model reliability), average variance extracted (AVE, ≥0.50 suggesting adequate convergence), and the square root of AVE (which should exceed the highest correlation with any other latent variable). If the AVE is below 0.50 but the CR exceeds 0.60, the convergent validity is still considered adequate [55].

Multiple linear regressions are conducted to estimate the relationships between sociodemographic, series preferences, personality traits, and the dimensions of binge-watching. The assumptions of multiple linear regression (the homogeneity of variance, the independence of observations, normality, and linearity) are met. Regression coefficients (R² and ΔR²), t-values from two-sided t-tests, p-values, and F change values are reported. The statistical significance level is set at 0.05. All the statistical analyses are performed using SPSS version 29 and AMOS version 29.

3. Results

3.1. Demographics

The sample includes 633 individuals, with 298 males (47.1%) and 335 females (52.9%). The participants’ ages ranged from 18 to 80, with a mean age of 40.44 years (SD = 15.15). In terms of marital status, 172 (27.2%) are single and not in a relationship, 109 (17.2%) are single but currently in a relationship, 270 (42.7%) are married or in a cohabiting relationship, 65 (10.3%) are divorced or separated, and 17 (2.7%) are widowed. Regarding their employment status, 400 (63.2%) are actively employed, 140 (22.1%) are students, 47 (7.4%) are retired, and 46 (7.3%) are unemployed.

3.2. Descriptives

Concerning the questionnaire about television series preferences, 456 (72.0%) participants watch action series, 399 (63.0%) drama, 383 (60.5%) comedy, 166 (26.2%) horror, 380 (60.0%) science fiction and fantasy; 338 (53.4%) romance, and 407 (64.3%) documentary series. In the total sample, 326 (51.5%) prefer calming series, and 307 (48.5%) prefer stimulating series. Men watch significantly more action [86.9% versus 58.8%; χ²(1) = 61.851; p < 0.001; Φ = −0.313], horror [41.6% versus 12.5%; χ²(1) = 68.902; p < 0.001; Φ = −0.330], and science fiction [783.5% versus 48.1%; χ²(1) = 42.507; p < 0.001; Φ = −0.259] series, that is, more stimulating [71.8% versus 27.8%; χ²(1) = 122.523; p < 0.001; Φ = −0.440] series than women; women watch significantly more dramas [70.7% versus 54.4%; χ²(1) = 18.168; p < 0.001; Φ = 0.169], comedies [74.9% versus 44.3%; χ²(1) = 61.919; p < 0.001; Φ = 0.313], romances [77.3% versus 26.5%; χ²(1) = 163.573; p < 0.001; Φ = 0.508], and documentaries [68.4% versus 59.7%; χ²(1) = 5.113; p = 0.024; Φ = 0.090], i.e., more calming [72.2% versus 28.2%; χ²(1) = 122.523; p < 0.001; Φ = −0.440] series than men (Figure 1).

3.3. Structural Analysis and Measurement Invariance Across Gender

Confirmatory factor analyses (CFAs) are conducted for the Portuguese sample to evaluate the fit of the seven-factor model derived from the original BWESQ validation [45]. The adequacy of the first-order, seven-factor model with 40 items, based on the preliminary BWESQ validation, is tested. As shown in

Table 1. Model’s fit indexes for the seven-factor model of the Binge-Watching Engagement and Symptoms Questionnaire (N = 633).

	N	χ2	df	χ2/df	RMSEA (CI)	CFI	IFI	SRMR
40 items, 7 factors first order, 1 factor second order	633	4831.036	736	6.564	0.094 (0.091–0.096)	0.892	0.892	0.048
40 items, 7 factors first order	633	4233.077	719	5.887	0.088 (0.085–0.091)	0.907	0.907	0.040
40 items, 7 factors first order, six correlations between errors	633	3546.655	702	5.052	0.080 (0.077–0.083)	0.925	0.925	0.037

Note. χ² = qui-squared; df = degrees of freedom; IFI = incremental fit index; CFI = comparative fit index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMS = standard root mean square.

, the goodness-of-fit indices for the BWESQ are acceptable after correlating the factors and establishing six error correlations based on modification indices.

The measurement invariance of the BWESQ across genders is reported in

Table 2. Multigroup CFAs of Binge-Watching Engagement and Symptoms Questionnaire according to gender (N = 633).

	χ2	df	χ2/df	RMSEA (CI)	CFI	IFI	SRMR	Comparisions	Δ RMSEA	Δ CFI	Δ SRMR
Configural invariance	5641.744	1424	3.962	0.069 (0.067–0.070)	0.886	0.887	0.027	NA	NA	NA	NA
Metric invariance	5698.563	1457	3.911	0.068 (0.066–0.070)	0.886	0.887	0.028	Configural vs. metric	0.001	0.000	0.001
Scalar invariance	5916.700	1497	3.952	0.068 (0.067–0.070)	0.886	0.887	0.038	Metric vs. scalar	0.000	0.000	0.010
Error invariance	6060.482	1525	3.974	0.069 (0.067–0.070)	0.886	0.887	0.053	Scalar vs. error variance	0.001	0.000	0.015

Note. χ² = qui-squared; df = degrees of freedom; IFI = incremental fit index; CFI = comparative fit index; RMSEA = root mean square error of approximation; CI = confidence interval; SRMS = standard root mean square; Δ RMSEA = change in the RMSEA compared with the previous model (expressed in absolute values); Δ CFI = change in the CFI compared with the previous model (expressed in absolute values); Δ SRMR = change in the SRMR compared with the previous model (expressed in absolute values). All models are significant at p < 0.001.

. The configural invariance by gender is confirmed during the first step of the multigroup CFAs. In the subsequent steps, the minor changes in the fit indices also support metric invariance by gender. Moreover, increasing the measurement constraints in the following steps do not lead to a significant deterioration in model fit, and error invariance across genders is achieved. This provides strong evidence that the BWESQ functions similarly for males and females. Most comparisons show changes of less than 0.01, further supporting varying levels of measurement equivalence between genders (Table 2).

3.4. Descriptives, Reliability, and Convergent and Discriminant Validity of BWESQ

Table 3. The correlations, Cronbach’s alpha, McDonald’s omega, composite reliability, average variance extracted (AVE), AVE square roots, mean and standard deviation of the Binge-Watching Engagement and Symptoms Questionnaire (N = 633).

	Pearson’s Correlations
	1	2	3	4	5	6	7	8	α	ω	CR	AVE	Mean (SD)
1. BWESQ total	0.86								0.99	0.99	0.98	0.75	2.31 (0.91)
2. Loss of control	0.958 **	0.91							0.98	0.98	0.97	0.83	2.11 (1.01)
3. Engagement	0.973 **	0.913 **	0.88						0.96	0.96	0.97	0.78	2.29 (0.94)
4. Dependency	0.959 **	0.958 **	0.908 **	0.94					0.97	0.97	0.97	0.88	2.12 (1.00)
5. Desire/savoring	0.900 **	0.776 **	0.873 **	0.797 **	0.88				0.94	0.94	0.95	0.77	2.61 (0.88)
6. Positive emotions	0.928 **	0.827 **	0.899 **	0.846 **	0.893 **	0.88			0.93	0.93	0.95	0.78	2.52 (0.90)
7. Binge-watching	0.974 **	0.949 **	0.933 **	0.941 **	0.843 **	0.874 **	0.91		0.96	0.96	0.97	0.82	2.26 (0.98)
8. Pleasure preservation	0.933 **	0.889 **	0.896 **	0.895 **	0.811 **	0.848 **	0.904 **	0.92	0.91	0.91	0.95	0.85	2.24 (1.00)

Note: ** p < 0.001; α = Cronbach’s alpha; ω = McDonald’s omega; CR = composite reliability; AVE = average variance extracted; bold (diagonal) = AVE square roots; SD = standard deviation.

presents the reliability indices for the total score and factors of the BWESQ. There are no notable differences between Cronbach’s alpha (α) and McDonald’s omega (ω), with most reliability values falling within the adequate-to-excellent range. Additionally, the composite reliability, average variance extracted (AVE), AVE’s square root, and the mean and standard deviation are calculated (Table 3). Almost all the values are within the acceptable reference range.

3.5. Descriptives, Reliability, and Convergent and Discriminant Validity of DDT

Reliability indices for the DDT total score and factors are displayed in

Table 4. The correlations, Cronbach’s alpha, McDonald’s omega, composite reliability, average variance extracted (AVE), AVE square roots, mean and standard deviation of the Dark Triad Dirty Dozen (N = 633).

Pearson’s Correlations
	1	2	3	4	α	ω	CR	AVE	Mean (SD)
1. DDT total	0.72				0.91	0.91	0.93	0.52	2.13 (0.76)
1 Machiavellianism	0.878 **	0.81			0.83	0.84	0.89	0.66	2.27 (0.88)
2 Psychopathy	0.811 **	0.585 **	0.85		0.87	0.87	0.91	0.73	1.79 (0.83)
3 Narcissism	0.866 **	0.660 **	0.521 **	0.85	0.88	0.88	0.91	0.73	2.31 (0.95)

Note: ** p < 0.001; α = Cronbach’s alpha; ω = McDonald’s omega; CR = composite reliability; AVE = average variance extracted; bold (diagonal) = AVE square roots; SD = standard deviation.

. Almost no differences between Cronbach’s alpha (α) and McDonald’s omega (ω) are observed, and most reliability values are adequate to excellent. Furthermore, the composite reliability, average variance extracted (AVE), square root of AVE, mean and standard deviation are calculated (Table 4), and almost all the values are also within the reference range.

3.6. Correlations, Skewness, Kurtosis, Tolerance, and Variance Inflation Factor of BWESQ and DDT

All the psychological variables are positively and significantly correlated ranging from 0.199 (between dependency and narcissism) and 0.526 (between dependency and psychopathy); in addition, all the variables have a normal distribution, but several BWESQ dimensions present multicollinearity (loss of control, engagement, dependency and binge-watching) (

Table 5. Correlations, skewness, kurtosis, tolerance, and variance inflation factor of BWESQ and DDT (N = 633).

					Pearson Correlations
	Skewness	Kurtosis	Tolerance	VIF	DDT Total	Machiavellianism	Psychopathy	Narcissism
1. BWESQ total	0.63	−0.64			0.425 **	0.339 **	0.518 **	0.245 **
2. Loss of control	0.72	−0.74	0.06	17.96	0.397 **	0.306 **	0.523 **	0.204 **
3. Engagement	0.54	−0.82	0.08	12.29	0.398 **	0.315 **	0.497 **	0.218 **
4. Dependency	0.73	−0.70	0.07	14.89	0.392 **	0.296 **	0.526 **	0.199 **
5. Desir/savoring	0.07	−0.84	0.17	6.06	0.401 **	0.346 **	0.407 **	0.273 **
6. Positive emotions	0.15	−0.82	0.14	7.41	0.418 **	0.341 **	0.465 **	0.271 **
7. Binge-watching	0.53	−0.83	0.06	16.50	0.423 **	0.333 **	0.516 **	0.247 **
8. Pleasure preservation	0.46	−0.94	0.15	6.76	0.410 **	0.324 **	0.491 **	0.242 **
Skewness					0.29	0.21	0.79	0.24
Kurtosis					−0.62	−0.74	−0.44	−0.75
Tolerance						0.49	0.63	0.54
VIF						2.06	1.60	1.87

Note: ** p < 0.001.

).

3.7. Hierarchical Regression Analyses Predicting Different Binge-Watching Dimensions

Hierarchical regression analyses are conducted to assess the sociodemographic, series preferences, and personality predictors for the BWESQ dimensions (

Table 6. Summary of the hierarchical regression analyses predict different binge-watching dimensions.

		β (t)
	BWESQ Total	Loss of Control	Engagement	Dependency	Desire/Savoring	Positive Emotions	Binge-Watching	Pleasure Preservation
Gender	−0.09 (−2.45) *	−0.13 (3.52) ***		−0.11 (−3.08) ***			−0.15 (−4.04) ***	−0.13 (−3.49) ***
Age	−0.15 (−3.71) ***		−0.19 (−4.53) ***		−0.26 (−5.85) ***	−0.19 (−4.46) ***	−0.14 (−3.32) ***	−0.19 (−4.66) ***
Employment	0.14 (3.68) ***	0.09 (2.86) **	0.15 (3.73) ***	0.11 (3.36) ***	0.15 (3.67) ***	0.12 (2.98) **	0.14 (3.55) ***	0.17 (4.37) ***
Action series	−0.07 (−2.06) **	−0.10 (−2.89) **		−0.10 (−2.87) **			−0.07 (−1.97) *	−0.10 (−2.74) **
Horror series	0.16 (4.28) ***	0.16 (4.34) ***	0.18 (4.82) ***	0.17 (4.61) ***	0.14 (3.46) ***	0.09 (2.33) **	0.17 (4.47) ***	0.15 (4.05) ***
Comedy series	−0.09 (−2.63) **	−0.07 (−2.02) *	−0.07 (−2.04) *	−0.11 (−3.18) **	−0.09 (−2.28) *	−0.08 (−2.18) **	−0.11 (−2.96) **	−0.09 (−2.51) *
Scientific fiction and fantasy series	0.16 (4.69) ***	0.17 (5.03) ***	0.15 (4.38) ***	0.15 (4.44) ***	0.13 (3.77) ***	0.14 (4.16) ***	0.14 (4.14) ***	0.14 (4.30) ***
Documentaries	−0.14 (−4.14) ***	−0.15 (4.48) ***	−0.12 (−3.45) ***	−0.16 (−4.84) ***	−0.10 (−2.81) **	−0.14 (−3.84) ***	−0.13 (−3.84) ***	−0.14 (−4.15) ***
Preferences: stimulating	0.106 (2.54) *	0.10 (2.46) *	0.10 (2.55) *	0.10 (2.44) *	0.11 (2.79) **	0.12 (2.88) **		0.13 (3.10) **
Machiavellianism	0.09 (2.41) *				0.18 (4.26) ***	0.13 (3.13) **	0.09 (2.30) *	0.09 (2.30) *
Psychopathy	0.24 (5.85) ***	0.32 (9.43) ***	0.30 (8.79) ***	0.32 (9.45) ***	0.11 (2.45) *	0.21 (4.67) ***	0.26 (6.15) ***	0.20 (4.79) ***
R2adj	0.44	0.43	0.41	0.45	0.35	0.36	0.41	0.44
F (2, 606)	40.55	88.94	77.24	89.23	26.73	35.76	43.78	29.46
p	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001	<0.001

β = standardized Beta; t = t test; *** = p < 0.001; ** = p < 0.010; * = p < 0.050.

). It is important to note that marital status, watching drama or romantic series, and narcissism do not contribute significantly to explaining any of the BWESQ dimensions.

Several factors predict different BWESQ dimensions. For the total BWESQ variance and pleasure preservation, being male, younger, active, not watching action series, comedy series, or documentaries, watching horror and science fiction/fantasy series, preferring stimulating series, and having high levels of Machiavellianism and psychopathy account for 44% of the variance. For loss of control, being male, active, not watching action series, comedy series, or documentaries, watching horror and science fiction/fantasy series, preferring stimulating series, and having high levels of psychopathy explain 43% of the variance. For engagement, being younger and active, not watching comedy series or documentaries, watching horror and science fiction/fantasy series, preferring stimulating series, and having high psychopathy values explain 41% of the variance. For dependency, being male, active, not watching action series, comedy series, or documentaries, watching horror and science fiction/fantasy series, preferring stimulating series, and having high psychopathy values account for 45% of the variance. For desire/savoring and positive emotions, being younger, active, not watching comedy series or documentaries, watching horror and science fiction/fantasy series, preferring stimulating series, and having high levels of Machiavellianism and psychopathy explain 35% of the desire/savoring variance and 36% of the positive emotions variance. Finally, for binge-watching, being male, younger, active, not watching action series, comedy series, or documentaries, watching horror and science fiction/fantasy series, preferring stimulating series, and having high levels of Machiavellianism and psychopathy explain 41% of the variance.

4. Discussion

This study aims to analyze whether sociodemographic characteristics, series preferences, and the dark triad of personality traits explain binge-watching behavior. To achieve this, the BWESQ is validated for the study population through a confirmatory factor analysis (CFA), and multigroup CFAs according to gender are conducted, testing four levels of measurement invariance: configural, metric, scalar, and error variance. To assess the model’s reliability, convergent and discriminant validity are evaluated using Cronbach’s alpha and McDonald’s omega coefficients, composite reliability, the AVE, and the square root of the AVE. Additionally, several multiple linear regressions are performed to estimate the relationships between sociodemographic factors, series preferences, personality traits, and binge-watching dimensions.

The model’s fit indices for the Portuguese version of the BWESQ are deemed acceptable after the factors are correlated and six error correlations are established [5], in line with the findings from the initial validation study by Flayelle et al. [45]. Furthermore, Flayelle et al. [56] validated the BWESQ across nine languages (English, French, Spanish, Italian, German, Hungarian, Persian, Arabic, and Chinese) to encourage international and cross-cultural binge-watching research. They report strong psychometric properties and fit in each language and support for equivalence across languages and gender. While Flayelle et al. [56] follow the recommendation of Kenny and McCoach [57], who argue that complex models with lower Tucker–Lewis index (TLI) and comparative fit index (CFI) values should not raise concerns if the RMSEA shows better fit, the RMSEA and the SRMR in our study, as well as in the studies by Flayelle et al. [3,45,58] fall below the thresholds of 0.08 and 0.10. Contrary to our study, Flayelle et al. [56] avoid applying any modifications based on modification indices to maintain the original factorial integrity of the scale and ensure comparability across countries.

Regarding reliability indices, our study observes no significant differences between Cronbach’s alpha (α) and McDonald’s omega (ω), with the majority of the reliability values being adequate to excellent. This finding aligns with Flayelle et al. [56], who also report minimal differences between ordinal Cronbach’s alpha and McDonald’s omega, with most reliability values being adequate to excellent, though with a few exceptions. In our study, the composite reliability, the average variance extracted (AVE), and the square root of the AVE are within the reference range, which is consistent with the findings of Flayelle et al. [56], who also report strong scale inter-correlations and convergent validity.

Our analysis achieves configural, metric, scalar, and error variance invariance, providing strong evidence that the BWESQ operates similarly for males and females. These results are also consistent with Flayelle et al. [56], who report measurement invariance for both language and gender, indicating that male and female TV series viewers interpret the BWESQ items conceptually similarly.

Concerning television series preferences, action series are the most watched, followed by documentaries, dramas, and comedies, with horror and romance series ranking lower. Approximately half of the sample prefers calming series, while the other half prefers stimulating series. These findings align with those of Flayelle et al. [58], who found a preference for high-energy narratives with acute stakes (i.e., action/adventure), though they also identified a preference for series with twisty plots and a tense atmosphere (i.e., horror), which contrasts with our results. Spangler [59] observed that while TV dramas remain the most popular among binge-watchers, nearly 25% of binge-watching is performed with other genres, like comedy and reality TV. This observation does not fully align with our findings, as men tend to watch significantly more action, horror, and science fiction series (i.e., more stimulating genres), while women prefer more tranquil genres such as drama, romance, comedy, and documentaries. These findings support the results of Starosta and Izydorczyk [8], who also found that women prefer comedies and dramas, while men watch fantasy or sci-fi series more frequently.

All the correlations between the BWESQ dimensions and the dark triad traits are positive and significant, indicating that higher binge-watching behavior is associated with higher levels of Machiavellianism, narcissism, and psychopathy. These results align with the findings of Starosta et al. [10], who observed that lower conscientiousness and low agreeableness, emotional stability, and intellect were linked to problematic binge-watching. Studies by Pittman and Steiner [60], Tóth-Király et al. [61], and Forte et al. [62] also reported that binge-watchers tend to be more neurotic, less agreeable, less conscientious, and less open to new experiences.

Interestingly, watching drama or romantic series and narcissism do not significantly contribute to explaining any BWESQ dimensions. This contrasts with the findings of Kircaburun and Griffiths [32] and Sheldon and Bryant [33], who associated narcissism with greater involvement in problematic social media use, online gaming, and internet use. Similarly, Jauk and Dieterich [39] and Eksi [40] linked narcissism to non-substance-related addictive behaviors. Furthermore, Castro et al. [63] reported that participants’ valence (emotional state) decreased after binge-watching, with negative effects slightly increasing after watching drama series, which is inconsistent with our finding that drama series were not a significant contributor to binge-watching behavior. The finding that narcissism does not predict binge-watching, while psychopathy and Machiavellianism do, likely reflects both theoretical and methodological nuances. Theoretically, narcissism prioritizes social validation and self-enhancement, which clash with binge-watching’s solitary, escapist nature. Psychopathy (impulsivity) and Machiavellianism (strategic goals) align better with binge-watching motivations. Methodologically, narcissism scales often conflate subtypes (e.g., grandiose vs. vulnerable), potentially missing escapist tendencies. Sample limitations (e.g., low narcissism variability) or statistical overlap with other traits (e.g., psychopathy) may also obscure its role. Future work could test narcissism subtypes or indirect pathways (e.g., social media use).

Psychopathy plays a significant role in explaining all the BWESQ dimensions, consistent with prior research linking psychopathy to problematic internet use and maladaptive coping strategies [32,37], as well as to non-substance-related addictive behaviors in both clinical and non-clinical populations [39]. Machiavellianism also significantly explains several dimensions of binge-watching (except for loss of control, engagement, and dependency). This finding aligns with studies associating Machiavellianism with the problematic use of social networks and online games, as individuals with high levels of Machiavellianism often become problematic internet users [35,36].

In our study, being male, younger, and active significantly contributes to explaining almost all the dimensions of binge-watching behavior. However, Merrill and Rubenking [14] found that the binge-watching duration was associated with being female, while Rubenking and Bracken [64] reported that neither age nor sex significantly explained binge-watching.

Not watching action series, comedy series, and documentaries contributes to explaining nearly all the dimensions of binge-watching, suggesting that these series may have a protective effect on binge-watching behavior. On the other hand, watching horror series, science fiction, fantasy series, and preferring stimulating series contribute to explaining binge-watching, suggesting that these genres may be more addictive than others. According to Ahmed [12], people who watch series frequently tend to engage in less problematic viewing, particularly when they watch foreign and dubbed dramas, action films, comedy programs, and documentaries.

This study presents a comprehensive analysis of binge-watching behavior, integrating sociodemographic factors, genre preferences, and the dark triad personality traits. While it offers valuable insights, several strengths and limitations warrant discussion. The study demonstrates methodological rigor through its use of a confirmatory factor analysis (CFA) and multigroup invariance testing, which robustly validate the Binge-Watching Engagement and Symptoms Questionnaire (BWESQ) for the Portuguese population. The inclusion of multiple reliability indices, such as Cronbach’s alpha and McDonald’s omega, alongside validity checks, like the average variance extracted (AVE) and composite reliability, strengthens confidence in the scale’s psychometric soundness. Another strength lies in the alignment of findings with Flayelle et al.’s international validation efforts, which underscores the BWESQ’s cross-cultural applicability. The establishment of measurement invariance across the genders further highlights the scale’s consistency in functioning similarly for males and females, enhancing its utility in diverse research contexts. Additionally, the study adopts an integrative approach by simultaneously examining sociodemographic factors, genre preferences, and personality traits, offering a holistic perspective on binge-watching predictors. The identification of genre-specific protective factors (e.g., action, comedy) and risk factors (e.g., horror, sci-fi) adds nuanced insights into how media consumption patterns may influence problematic viewing behaviors.

However, this study has several limitations. Its focus on a Portuguese sample restricts its generalizability to other cultural or linguistic contexts, and conflicting findings—such as discrepancies in gender/age effects compared to prior studies like Merrill and Rubenking’s—may reflect unexamined cultural or sampling differences. While modifications to the CFA model improved the fit indices through error correlations, this approach risks overfitting the data, potentially compromising the BWESQ’s factorial integrity and comparability to Flayelle et al.’s unmodified version. The cross-sectional design also limits causal interpretations, as correlations between dark triad traits and binge-watching could stem from shared underlying factors, like impulsivity, rather than indicating direct causation. Longitudinal or experimental designs would help clarify these relationships.

Another limitation includes a higher prevalence of active workers compared to students and a mean age of 40.44 years, which may influence series consumption patterns compared to the younger populations reported in the literature. Additionally, as the instruments are self-reported, answers may be influenced by social desirability biases. A key limitation is the lack of existing research on the relationship between binge-watching and dark triad traits.

Notably, the non-significant role of narcissism in explaining binge-watching contrasts with its established links to other addictive behaviors (e.g., social media use), raising unanswered questions about the unique psychological mechanisms driving binge-watching. The study’s exclusive focus on the dark triad, rather than broader personality frameworks like the Big Five, further narrows its scope, overlooking traits, such as low conscientiousness or emotional instability, that prior research associates with problematic viewing. Reliance on self-reported data introduces potential biases, particularly for sensitive traits like psychopathy, and triangulation with behavioral metrics could strengthen findings. Mechanistically, the study does not explore why specific genres correlate with binge-watching, leaving open whether narrative structures (e.g., cliffhangers), emotional arousal, or viewer identification drive these associations.

Theoretically, the study advances the understanding of binge-watching’s psychosocial correlates, but the practical implications remain unclear. For instance, how genre preferences might inform interventions for problematic viewing or how protective genres like documentaries could be leveraged in harm-reduction strategies are underexplored. Future research should address these gaps through diverse samples, experimental designs, and integrative frameworks to disentangle the complex drivers of binge-watching behavior.

5. Conclusions

The findings underscore a robust link between binge-watching behavior and specific personality traits, particularly psychopathy and Machiavellianism, which emerge as central predictors of excessive viewing. These traits, characterized by manipulative tendencies and emotional detachment, appear to drive maladaptive engagement with the media. Sociodemographic factors—such as male gender, younger age, and higher activity levels—further amplify binge-watching tendencies, while genre preferences play a dual role: stimulating genres like horror, science fiction, and fantasy correlate with increased risk, whereas action, comedy, and documentaries may mitigate problematic consumption.

By validating the BWESQ cross-culturally and elucidating the dark triad’s influence, this study advances media psychology research. However, unresolved contradictions—such as inconsistent gender/age effects and the negligible role of narcissism—highlight gaps in understanding binge-watching’s psychological underpinnings. Methodological constraints, including the reliance on self-reports and the cross-sectional design, limit causal interpretations.

Future research should prioritize the cross-cultural validation of the BWESQ and longitudinal designs to clarify the causal links between dark triad traits and binge-watching. Expanding personality frameworks (e.g., Big Five) could resolve contradictions like narcissism’s limited role here versus its impact on other addictions. Integrating behavioral data (e.g., streaming analytics) with the self-reported data would reduce the bias. Mechanistic studies should explore why stimulating genres (horror, sci-fi) heighten risk, while protective genres (comedies, documentaries) may promote healthier habits. Clinically, tailoring interventions for high-risk groups (e.g., younger males) could be achieved by addressing traits like psychopathy and leveraging genre preferences. Refining the BWESQ for younger, diverse samples and standardizing the methodological adjustments (e.g., CFA error correlations) would ensure cross-study comparability. Public health efforts should advocate for ethical platform design (e.g., auto-pauses) and awareness campaigns about genre-specific risks. Finally, interdisciplinary collaboration is key to balancing mindful media engagement with the cultural value of storytelling. Future investigations should employ diverse, representative samples and experimental approaches to unravel the mechanisms linking narrative structures, emotional arousal, and viewer behavior.