A total of 301 respondents voluntarily completed the online survey, and one participant was excluded because s/he was not enrolled in a healthcare curriculum. Therefore, the final sample comprised 300 participants (82% women; M_age = 23.5; SD = 5.97). Most were Bachelor’s students (n = 215), with Master’s students representing 85 participants, and mainly came from nursing (35%), physiotherapy (26%), psychology (9%), and medicine (8%) departments (see Table A in Nasello, 2025 for descriptive results). Most participants identified as Caucasians (n = 265) and reported an average parental socioeconomic status (64%). The financially independent students (n = 51, 39%) reported a low socioeconomic status (income < €1200/month).

Inclusion criteria required enrollment in a healthcare program, completion of at least one clinical internship, and age ≥ 18 years. Ethical approval was obtained from the Ethics Committee of the Department of Psychology at the University of Liège (Belgium), Reference N°: 2122-040. Informed consent was obtained online.

A power analysis conducted with G*Power 3.1.9.7 (Faul et al., 2007) determined that we would require at least 167 participants for our main analysis to achieve a power of 0.80 with an α-error settled at 0.05, 8 degrees of freedom, and to detect moderate effect sizes (settled at 0.3). The online data file is available at Nasello (2025).

Structural equation modeling (SEM) was used to examine the relationships between empathy, WTH, and sociodemographic variables (gender, age, and grade). The model’s goodness of fit was assessed using five statistical indices: the chi-square goodness-of-fit (χ²), the Root Mean Square Error of Approximation (RMSEA), the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), and the Standardized Root Mean Square Residual (SRMR). A non-significant chi-square value indicates a satisfactory fit of the model. However, since the chi-square test is influenced by sample size and can yield significant results with larger samples, a χ²/df ratio is used. The χ²/df ratio should not exceed 5 to display good goodness of fit. RMSEA, CFI, TLI, and SRMR were used to confirm the model’s fit (Schumacker & Lomax, 2010). CFI and TLI values greater than .9 indicate an adequate fit, while a value greater than .95 suggests a good fit. For the RMSEA and SRMR, a value below .08 indicates an adequate fit, while a value below .05 represents a good fit (Hu & Bentler, 1999).

A t-paired test and a correlation matrix accompanied the main analysis (SEM). The analyses were conducted using JAMOVI computer software, Version 1.6.23 (Jamovi Project, 2019) and R, Version 4.2.2 (R Core Team, 2021), lavaan package 0.6.15 (Rosseel, 2012).

Table 1 displays the descriptive statistics for all variables. A paired-samples t-test showed that WTH differed significantly across conditions, t₍₂₉₉₎ = -5.62, p < .001, d = -.32: future healthcare professionals reported higher WTH toward patients who had engaged in healthy behaviors compared to those with unhealthy behaviors (small effect size).

The correlation matrix (see Table 2) revealed that affective empathy was significantly correlated to WTH_H, r = .17; p < .01, WTH_UH, r = .18; p < .01, and WTH_Tot., r = .18; p < .01. Cognitive empathy was also significantly correlated to WTH_H, r = .16; p < .01, but the correlation was stronger for WTH_UH, r = .23; p < .001, and WTH_Tot., r = .20; p < .01. Overall, the correlations between empathy and WTH domains were small.

Regarding the sociodemographic variables, gender (women were coded as 1) was significantly correlated to affective empathy, r = .34; p < .001, WTH_UH, r = .14; p < .05, WTH_H, r = .14; p < .05, and WTH_Tot., r = .13; p < .05. Grade (Bachelors were coded as 1) was also significantly correlated to cognitive empathy, r = -.17; p < .01, and WTH_H, r = .12; p < .05. Finally, age was not significantly associated with empathy or WTH domains, ps > .05. All these correlations were small, except for the relationship between gender and affective empathy that is moderate.

The final model showed adequate goodness of fit indices, χ²/df = 2.71, CFI = .97, TLI = .95, RMSEA = .075, SRMR = .038. In this final model, we set the variance of WTH_UH at 1. As shown in Figure 1, empathy had a (moderate) significant direct link with WTH, β = .29; p < .005 (see Table D; Nasello, 2025), confirming the main hypothesis of this study. Gender, age, and grade did not significantly impact WTH, ps > .05.

Regarding empathy, we found that gender and grade had a significant direct link with empathy (gender: β = .40, p < .001; grade: β = -.17, p < .05; see Table D; Nasello, 2025). In other words, women presented higher levels of empathy than men (the effect size of this association is moderate) and Master’s students had higher empathy levels than Bachelor’s students (the effect size is small). Refer to Tables D–G in Nasello (2025) for complementary descriptive statistics in the Supplementary Materials.

Regarding sociodemographic variables such as age, gender, or grade, none of these variables had a significant influence on WTH. This suggests that, across different levels of training, ages, and genders, future healthcare professionals express a similar baseline willingness to help. These results replicate the nonsignificant associations reported by Hajdu et al. (2022).

Regarding the effects of these sociodemographic variables on empathy, we found a significant and positive effect of gender: women exhibit higher levels of affective empathy. Additionally, we observed a significant negative effect of the grade variable: master’s students have higher levels of empathy.

The former result is a well-established empirical finding (Baron-Cohen & Wheelwright, 2004; Jolliffe & Farrington, 2006; Nasello et al., 2018; Nasello & Triffaux, 2020; Triffaux et al., 2019; Triffaux et al., 2023). Sex role stereotypes have been suggested as a potential explanation for the gender effect on empathy (Hoffman, 1977), but other authors argue that these gender differences are also present in elementary forms of empathy, such as mimicry, indicating a deeper-rooted effect (Goldman & Sripada, 2005; Hermans et al., 2006; Lakin et al., 2008; Schulte-Rüther et al., 2008).

The latter result may seem surprising because several studies have demonstrated a significant decline in empathy scores during education, particularly in the field of medicine (Hojat et al., 2004; Triffaux et al., 2019; Ward et al., 2012), as well as in various healthcare fields, including medicine (Nunes et al., 2011). However, the grade effect on empathy aligns with a recent study that revealed higher empathy scores among healthcare students (specifically, medical students) during the COVID-19 pandemic (Triffaux et al., 2023). The authors observed significantly higher scores in affective and cognitive empathy in the pandemic cohort compared to two pre-pandemic cohorts, using two different empathy measurement tools. The authors proposed that these higher scores observed in the pandemic cohort could be either transient or permanent. In either case, these higher scores could potentially have a detrimental effect (refer to Triffaux et al., 2023 for more details), sustaining the need to intervene at the medical curriculum level by proposing targeted courses on empathic skills.

The study’s findings echo existing calls for structured training interventions into the education of future healthcare professionals to cultivate appropriate empathy skills (Hojat, 2016; Kilic et al., 2021; Nasello & Triffaux, 2020; Nasello & Triffaux, 2023; Riess & Kraft-Todd, 2014; Triffaux et al., 2019; Triffaux et al., 2023; Zhu et al., 2021). Such tailored interventions should prioritize the development of effective emotion regulation skills and the promotion of specific facets of empathy, such as empathic concern and perspective-taking (Nasello & Triffaux, 2023; Triffaux et al., 2023).

There are several ways to improve empathic skills. One model, proposed by Riess and Kraft-Todd (2014), is called E.M.P.A.T.H.Y. It focuses on nonverbal empathic skills and includes the following components:

E: Eye contact.

Alongside proposals aimed at developing relational skills centered on empathy, other approaches could be suggested and adapted for (future) healthcare professionals. For instance, authors such as Bianco and Castelli (2023), as well as Bateman and Fonagy (2019), have put forward various models intended to foster theory of mind and mentalizing capacities. Although these models have not yet been applied with the explicit goal of enhancing the interpersonal abilities of (future) healthcare professionals, they show considerable promise, particularly due to their clear proximity to cognitive empathy.

Bianco and Castelli (2023), for example, have identified three key components for cultivating theory of mind skills (i.e., the ability to attribute mental states to ourselves and others in order to make sense of people’s social behavior). These components include:

These three components can be meaningfully linked to the framework proposed by Bateman and Fonagy (2019), who highlight how neural dynamics jointly enable the mind to:

The present study demonstrates that future healthcare professionals’ willingness to help varies depending on patients’ prior health behaviors, with lower WTH observed toward patients who engaged in unhealthy behaviors. Additionally, the findings provided strong evidence of empathy playing an increasing role, with a moderate effect size, in shaping the WTH among future healthcare professionals. Hence, our findings suggest that the Theory of Planned Behavior and Reasoned Action may require refinement. Indeed, attitudes and subjective norms will have a significant impact on the intention of future caregivers to help and care for a patient, as will perceived behavioral control and conscious decision-making. However, on the one hand, we see that certain individual characteristics of these future caregivers (their empathy) also play a significant role in their intentions to help, and on the other hand, patients’ prior health behaviors may influence how they receive care.

Overall, our results advocate for the promotion of empathic skills and related aspects within the academic curriculum of future healthcare professionals (e.g., incorporating targeted courses on empathic or mentalization skills). Addressing this challenge in education is crucial and ranks among the most fundamental tasks for the forthcoming decades.

The study’s main limitation lies in the fact that we measure WTH but not the actual helping behavior itself. Although it is likely that WTH translates into helping behaviors, this point remains hypothetical.