Participants

A convenience sample was recruited via social media, listservs, and word of mouth, targeting students enrolled at two nearby U.S. universities. Participants were recruited between July 11^th and August 1^st, 2023. The preregistered exclusion criteria were: (a) not attending a U.S. university (self-reported), (b) being outside the age range of 18–24 years, (c) not identifying as either a woman or a man (due to the binary gender-specific BD standards applied in the analysis), and (d) incomplete survey responses. The final sample comprised N = 116 participants (women = 73.28%, age range 18–23, M_age = 19.93, SD_age = 1.24).

Procedure

The study was conducted entirely online via the online survey platform Qualtrics, accessible on participants’ personal computers or mobile devices. After providing informed consent, participants reported demographic information including age, gender, and race/ethnicity. Recruitment targeted students at the University of North Carolina at Chapel Hill and North Carolina State University; however, institutional affiliation was not collected to protect participant privacy. Next, the participants were asked questions regarding their drinking behaviors, followed by a questionnaire measuring affective and cognitive empathy. Upon completion, participants received a full debriefing via a document implemented within Qualtrics. The study took approximately 20 minutes to complete.

Measures

Patterns of alcohol consumption

Patterns of alcohol consumption were evaluated using two self-report measures. In accordance with the definition of BD provided by the National Institute on Alcohol Abuse and Alcoholism ([6],also, [7]), BD was defined as five or more drinks in two hours for men and four or more drinks in two hours for women (NIAAA, [6]). EBD was defined as consuming ten or more drinks in one sitting for men and eight or more drinks in one sitting for women [5]. To be classified into the BD or EBD group, participants were required to have engaged in the respective behavior at least twice within the past six months (in line with [7], for review). The non-BD group consisted of those who did not meet the criteria for either BD or EBD. Individuals who met the EBD criteria were assigned to the EBD group and excluded from the BD group. Therefore, the three drinking groups (non-BD, BD, and EBD) were constructed to be mutually exclusive to ensure independence.

The second measure used for the continuous variable of hazardous alcohol consumption was the Alcohol Use Disorders Identification Test – Consumption (AUDIT-C) [40], which assesses alcohol consumption frequency, intensity, and the frequency of consuming more than six drinks on a single occasion. The AUDIT-C is a reliable measure for assessing problematic alcohol consumption in adolescents [41] and college students [42], and has been validated in both online and in-clinic modalities [43].

Empathy questionnaire

The Questionnaire of Cognitive and Affective Empathy (QCAE) is a measure of trait empathy comprising two subscales: cognitive empathy (19 items) and affective empathy (12 items) [44]. Participants rated the extent to which each statement described them on a 4-point scale ranging from 1 (strongly disagree) to 4 (strongly agree). Higher scores on the respective subscales indicate higher levels of empathic ability. The QCAE [44] has been found to have both good reliability and validity in both self-report surveys and clinical samples [44,45]. In the current sample it demonstrated adequate internal consistency (cognitive empathy α = 0.82, affective empathy scale α = 0.77).

Attention check

Given the online format of the study, an attention check item was implemented within the QCAE [44] to ensure data quality by verifying that participants read the items carefully. The item read: “Please select ‘strongly disagree’ to demonstrate that you are paying attention to this question”.

Preregistration

The study design and data analysis were preregistered: https://aspredicted.org/Q89_ZSN.

Inclusivity in global research

The study was conducted in accordance with the Declaration of Helsinki (1964 and its later amendments) and approved by the Ethical Board of the University of Vienna, Faculty of Psychology (EK reference number 00577, amendment 16 to the project 00412). Prior to study participation, participants were presented with an information sheet outlining the study’s purpose, procedures, and their rights as research participants. Informed consent was obtained electronically, with participants indicating agreement by selecting ‘I agree.’ This confirmed their understanding of the study, that their participation was voluntary and uncompensated, and their right to withdraw at any time without penalty. All responses were collected in an anonymized format.

Additional details on the ethical, cultural and scientific considerations related to inclusivity in global research are provided in the Supporting Information (S1 File).

Data analysis

Analyses were conducted using JASP [46], (ver. 0.18.3) and R [47], (ver. 4.4.0). The data analysis followed the preregistered plan, and the full dataset is available online [48].

First, we conducted correlation analyses to examine the relationships among drinking group (three levels: non-BD, BD, and EBD), hazardous drinking (AUDIT-C; [40]), cognitive and affective empathy (QCAE; [44]), and demographic variables (age, gender).

Next, we conducted analyses with drinking group as a factor (non-BD, BD, and EBD), testing the effects of cognitive and affective empathy using a multinomial regression analysis, followed by a one-way analysis of variance (ANOVA). We then used a linear regression model to assess the effects of empathy on hazardous alcohol consumption (AUDIT-C).

For the multinomial regression analysis, the dependent variable was the drinking group (non-BD, BD, EBD) with the non-BD group serving as the reference group. The independent variables were cognitive empathy, affective empathy, and the demographic variables (age, gender; see Equation 1).

(1)

We then conducted ANOVAs to test whether there were differences in cognitive and affective empathy across the three drinking groups.

To examine the relationship between empathy and hazardous drinking, we conducted a linear regression analysis with cognitive empathy, affective empathy, and control variables (age, gender) as the independent variables and the AUDIT-C scores as the dependent variable (see Equation 2).

(2)

For all analyses performed, the significance level was set at p < .05.

Sample characteristics

Out of 128 participants, six participants were excluded for failing the attention check. In addition to preregistered exclusions, we applied two data-quality/outlier procedures not specified a priori. First, to address potential careless responding, we calculated Mahalanobis distance for 31 of our QCAE items using the R package “careless” [49]. Three participants were excluded as multivariate outliers, with Mahalanobis distances exceeding the χ² cutoff at the 99% confidence level (p < .01). Second, three additional participants were excluded because their empathy scores (cognitive or affective, [44]) or AUDIT-C scores [40] deviated by more than ±2.5 SD from the group mean, indicating extreme cases.

The final sample comprised N = 116 participants ( 73.28% women), aged 18-23 years (M_age = 19.93, SD_age = 1.24). An a priori power analysis conducted using G*power [50] indicated that a minimum sample of 100 participants would provide 80% power to detect a medium-to-large effect size (f = .32) in a one-way ANOVA at α = .05 with three groups. Most participants identified as White/Caucasian (69.83%). Of the total sample, 46.55% were in the non-BD group (n = 54), 32.76% of participants were in the BD group (n = 38), and 20.69% of participants were in the EBD group (n = 24). The mean AUDIT-C score across participants was 3.61 (SD = 2.63). See Table 1 for detailed sample characteristics.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive Statistics for the total sample and drinking groups.

https://doi.org/10.1371/journal.pone.0341842.t001

As shown in Table 2, men (n = 31) had a mean AUDIT-C score of 4.84 (SD = 2.83), whereas women (n = 85) had a mean score of 3.17 (SD = 2.41). A score of 3 or more for women and 4 or more for men is considered hazardous ([51], but see [52] for calibrated cutoffs). Thus, on average, both men and women in this sample exceeded the gender-specific thresholds for hazardous drinking.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Descriptive statistics of the sample by gender.

https://doi.org/10.1371/journal.pone.0341842.t002

Statistical analyses

To obtain a complete overview of the relationships among the key study variables, we computed Spearman’s rho correlations (appropriate given the ordinal nature of some variables) between drinking group (with three levels, non-BD = 0, BD = 1, EBD = 2), AUDIT-C scores, affective and cognitive empathy, age (18-23), and gender (coded: 1 = man, 2 = woman). The results are presented in Table 3.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Correlation matrix.

https://doi.org/10.1371/journal.pone.0341842.t003

A strong positive correlation between drinking group (non-BD, BD, and EBD) and AUDIT-C scores was found (r_s = .80, p < .001), indicating that these two measures (group categorization and AUDIT-C scores) align in their assessment of drinking behavior. Gender (coded 1 = man, 2 = woman) was negatively correlated with AUDIT-C scores (r_s = −.27, p = .004) and with the drinking group (r_s = −.22, p = .019), indicating more problematic overall drinking behavior in men. Gender was positively correlated with affective empathy (r_s = .40, p < .001), suggesting that women tend to show higher affective empathy scores. Finally, age significantly correlated with AUDIT-C scores (r_s = .23, p = .01), such that higher age was associated with higher hazardous drinking scores. Fig 1 presents mean scores on the AUDIT-C, QCAE cognitive empathy, and QCAE affective empathy, separated by drinking group.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. AUDIT-C, QCAE Cognitive Empathy, and QCAE Affective Empathy scores.

AUDIT-C [40], QCAE Cognitive Empathy, and QCAE Affective Empathy [44] scores, separated by drinking group (non-binge drinking [NBD], binge drinking [BD], and extreme binge drinking [EBD]).

https://doi.org/10.1371/journal.pone.0341842.g001

To examine the relative contributions of cognitive and affective empathy to drinking group membership, we conducted a multinomial logistic regression analysis, controlling for age and gender. The overall model was not significant (p = .24).

Similarly, one-way ANOVAs with drinking group (non-BD, BD, and EBD) as the factor revealed no significant group differences in cognitive empathy (p = .68) or affective empathy (p = .63).

Finally, a regression analysis indicated that neither cognitive (p = .78) nor affective empathy (p = .33) significantly predicted AUDIT-C scores when controlling for age (p = .03) and gender (p = .002).

Limitations

Several limitations must be acknowledged. First, the cross-sectional design precludes causal inference; for example, it remains unclear whether the observed positive correlation between hazardous drinking and age indicates increased risk with advancing age. Future research should also employ longitudinal designs to examine whether–and under what conditions– both stable (trait) and situational (state) aspects of empathy relate to the development and progression of binge drinking [e.g., 39].

Second, the sample comprised exclusively college students, who may experience unique pressures and motivations for drinking, which limits the generalizability of the findings beyond college populations. Relative to their non-college peers, college students’ alcohol use is more strongly influenced by perceived peer norms, suggesting that the college environment may amplify the impact of social influences on drinking behavior [61]. College students also represent a unique population in which to study empathy, as prior research using the IRI has documented gradual declines in empathy across cohorts, with more recent generations scoring lower on the subscales Empathic Concern and Perspective Taking [62, for a meta-analysis].

Further, because participation was voluntary, self-selection bias may have inflated the observed prevalence of alcohol use in this sample, as students with more regular or heavier drinking patterns may have been more likely to participate. Lastly, reliance on self-report measures for both alcohol consumption and empathy may have introduced bias, such as underreporting of drinking behaviors or socially desirable responding on empathy items. As it pertains to alcohol consumption, college students often misjudge standard drink volumes and tend to overestimate the appropriate amounts, leading them to overpour drinks and consequently underreport their actual drinking levels [63]. This raises the possibility that rates at or above BD thresholds were underreported in the present investigation [(see also 64)]. Taken together, the reliance on self-report measures, a convenience sample drawn from only two universities, and the voluntary nature of participation raise the possibility that the nonsignificant findings may reflect characteristics of this sample rather than a true absence of association. Accordingly, the results should be interpreted with caution and warrant further investigation.

Future directions

Future research should seek to replicate these findings across diverse populations and age cohorts using probability-based sampling, ensure balanced gender representation, and include non-college comparison groups. Additional variables of interest that may influence alcohol use include self-efficacy in resisting peer pressure [e.g., 24], popularity, and familial alcohol consumption. Previous research suggests that adolescents with a family history of alcohol misuse are more likely to exhibit alcohol-related problems [65, 66, 67, see 7 for review]. However, more recent work has begun to question the strength of this association, with Tschorn et al. [68] proposing that familial risk factors may be less influential than previously thought. Instead, social and personality factors, such as impulsivity and risk-taking, appear to play a more substantial role in the development of alcohol misuse. In addition to these influences, prior work has linked attention deficit hyperactivity disorder (ADHD) to binge drinking among college students [69]. Notably, adults with subclinical ADHD have been found to exhibit impairments in affective, but not cognitive, empathy [70], highlighting a potential pathway linking empathy-related processes to problematic drinking.

Looking ahead, future studies may benefit from incorporating experimental paradigms to assess empathy (e.g., [71,72]), as evidence from experimental pain research suggests an additional mechanism linking alcohol use and pain empathy. Alcohol exerts reliable analgesic effects [for a meta-analysis, 73], with animal models indicating that these effects are mediated through opioid-related pathways [74]. Importantly, studies manipulating pain perception through placebo-analgesia show that reducing one’s own pain diminishes empathic responses to others’ pain [75, see 76 for review]. Similarly, acetaminophen (paracetamol) has been found to blunt empathic responses to both others’ physical and social pain [77]. Extending this line of research, it is important to examine whether alcohol’s opioid-related analgesic properties differentially impact empathic responding – for example, by diminishing empathy for pain more strongly than for other domains, such as empathy for unpleasant touch. This idea is informed by Rütgen et al. [78], who found that placebo analgesia reduced empathy for both pain and unpleasant touch, indicating that the effect extends, at least in part, to unpleasant affect; however, only pain empathy was blocked by an opioid antagonist, providing evidence that opioid-related mechanisms specifically underlie empathy for pain. Consequently, research on alcohol’s social effects should distinguish between empathy domains and explore the possibility that alcohol and pharmacological analgesics act through overlapping neurochemical pathways to influence empathic behavior.

Finally, manipulations of pain perception also have real-life consequences. Hartmann et al. [79] showed that placebo analgesia not only reduces empathy for pain but also diminishes prosocial behavior, such as the willingness to exert effort to help. Such findings raise the possibility of a vicious cycle in which alcohol consumption dampens empathy and prosocial behavior, thereby reducing reciprocal social support and ultimately perpetuating a spiral that exacerbates alcohol-related problems. This reinforces the broader societal implications of alcohol’s effects on empathy, and the need to consider both neural mechanisms and social dynamics when addressing alcohol misuse.