According to Jonasson and Thiborg [5], a sport is a regulated, competitive and physical game [6]. However, this definition has been criticized for being incomplete, given that the criteria of institutionalization [7] and social acceptance [6] were not considered.

As Parry [8] points out, athletic physical activity is not defined by a certain intensity but by whether the action affects the final result or not. This is also the criterion followed by the International Olympic Committee, and it also seems applicable to esports [8]. Institutionalization, on the other hand, is composed of the universalization of the rules and promotion of the sport itself [7]. The creation of common rules is hampered by the fact that the video games on which esports are based have extremely short life spans [9]. To compensate for this problem, esports are progressively moving from a “bottom-up” regulation, that is, from fans and members of the video game community in question to a “top-down” one, where the manufacturer of the game itself, together with its release, is also involved in the organization of its esport scene. Lastly, the social recognition of esports is being affirmed on several levels [6]: on the one hand, from the players, who started defining themselves as athletes since the first video game tournaments in the ‘80s [9]; on the other, from the spectators, whose main motivation to observe this kind of sport is linked to the high competitiveness put on the field by the athletes, similarly to traditional sports’ audiences [10]. Therefore, considering these characteristics, it seems proper to define sport as a physical, competitive, institutionalized, and socially accepted activity.

Eplayers can start their careers at a very young age, i.e., as early as 14-15 years old [11], and a sharp decline can already be seen from 24-25 years onwards [12]. Beyond the economic aspect, esporters differ from other people playing video games purely for fun due to the fact that they consider playing video games competitively a job, and engaging in the game differently [13]. The esports market could be considered as a superstar market, characterized by the fact that, similarly to what happens in the traditional sports and entertainment industry, players’ salary and prestige increase exponentially, and not linearly, as the players’ skills and results increase [14].

In the literature, it is highlighted that, for esporters’ careers, spending many hours a day in contact with electronic devices is linked to various physical and psychological consequences.

Physical ones can be grouped into four macro categories:

• Visual fatigue problems: e.g., Computer Vision Syndrome (Visual fatigue, low back pain, and tension headaches are the most common symptoms [15].
• Musculoskeletal problems: Caused by maintaining a sitting posture for several hours a day [16].
• Cardiovascular problems: While the sedentary life of athletes seems to negatively impact blood circulation, on the other hand, the variability of heart rate increases if one competes in esports, improving the body’s ability to adapt to different situations [2].
• Injury: Another element in common with traditional sports, in the esports field, the most common complications are related to pain in hand or wrist [15].

The consequences on mental health, instead, can be related to:

• Stress: Linked to increases in blood pressure, respiratory rate, and skin temperature [2].
• Burnout: In the context of esports, it seems to be correlated with the personality factor of agreeableness, probably given its incompatibility with the particularly aggressive and competitive environment that characterizes esports [17].
• Addiction: Characterized by three key symptoms, i.e., withdrawal, craving, and tolerance. Internet Gaming Disorder (IGD) is the only behavioral addiction, except for gambling addiction, in all of DSM-5. IGD addiction is defined as “persistent and recurring use of the internet to participate in games, often with other players, leading to clinically significant impairment or distress” (DSM-5, pp. 795-798; American Psychiatric Association, 2013), but this conceptualization has been criticized for failing to adequately distinguish between a strong passion for video games and a real addiction [18].

Shifting the focus on the positive aspects, much more related to the sport activity, eplayers can show a number of skills: the ability to interface with artificial intelligence and electronic devices; communication skills; improved cognitive abilities (such as increased visual perception of complex stimuli, better attentional resources, reduced response times, etc. and the ability to endure a long sitting position without compromising performance [11]. Furthermore, it has been observed that some eplayers habitually employ some psychological techniques during their gaming sessions in order to increase their performance [19].

A number of studies have shown that psychological training skills can lead to improved esports performance. For example, psychological skill training (PST) to improve some mental techniques such as goal setting, self-talk, mental imagery, mental rehearsal and relaxation seem to have positive effects on team cohesion and performance [20, 21] and even manuals about psychological training for esporters are also being developed [22]. More generally, it has been observed that higher psychological skills and higher self-regulation and social support are linked to better esport performance [23]. Psychological training techniques are habitually employed in traditional sports psychology, such as those focused on mental toughness and stress-coping processes, having a beneficial effect in this regard, also for esporters [24].

In light of these findings, this paper will be focused on the development of an instrument useful to sports psychologists to monitor some psychological aspects which could enhance esport performance, based on an Italian model of psychological training, already widely used in traditional sports practice, adapting it to the particular features of esports.

The present research is based on the “S.F.E.R.A. model”, developed by Vercelli [25, 26]. It is widely used in sports practice, adopted by the Italian National Olympic Committee (CONI), by several Olympic athletes and even by some professional clubs of various sports. According to the model, there are five mental factors to train to improve performance.

• Synchrony: The ability of being fully present in what one is doing at the moment of performance, supported by techniques such as self-talk and self-observation. It is the key to reaching all one's resources, especially the unconscious ones, and making them available to the performance. In fact, it is common for athletes who have reached this state to report that they move almost automatically, as if they were not driving their own bodies.
• Strength: To bring only and exclusively the athlete's strengths into the competition (physical, technical, and psychological skills and abilities that the athlete acknowledges possessing in an excellent way). It is linked to the concept of self-efficacy. Any defects should be temporarily forgotten. It is important to emphasize that this attitude is functional only at the moment of the game [27], while during training, athletes should work to their limits.
• Energy: The active use and regulation of strength and power, which allows athletes to best express their resources in a self-controlled way. It is further divided into Dynamism (enthusiasm and be ready to face unexpected events) and Dominance (assertiveness on opponents and influence on teammates).
• Rhythm: The orderly succession of time intervals, which generates the right flow in the sequence of movements and gives quality to the action.
• Activation: A motivational engine of an emotional and passionate nature, favored by connection with one's emotions, self-talk, and self-analysis. Activation is, more than any other factor, an internal sensation, the athletes’ belief that they are ready to perform in the match.

To achieve maximum performance, there must be a strengthening and a total balance between these five factors. Furthermore, the mind and body, conscious and unconscious, must work in unison, as if they were a unity, clearing the mind of any other thought.

Subsequently, this theory was enriched by qualitative methods of analysis and graphical visualization of the five factors, such as the S.F.E.R.A. mandala [25, 27] and quantitative methods, such as various questionnaires suited to different contexts of sport life: individual activity, group sports, training and competition, etc. [28].

Some similarities are noticeable between the theoretical descriptions of such factors and several classical constructs more common in sport psychology. For example, the pleasant dimension of Activation and the automaticity of thought and movement described in Synchrony closely resemble the dimension of Flow at Sport, and especially its subdimension of Absorption [29]. On the other hand, the focus on effort and physicality of Rhythm and Energy reminds us of the construct of Engagement, and of its Vigor subdimension in particular [30]. Thus, considering the use of the S.F.E.R.A. model applied in the esports’ field, it seems possible to find relations between Flow and Synchrony with Activation, and between Engagement and Rhythm with Energy, other than Activation again, as also the construct of Engagement includes an emotional aspect [30]. Moreover, as the five factors of S.F.E.R.A. sustain the healthy development of one's passion, a negative or absent relation with addiction is expected.

Consequently, the present research pursues the following goals:

• To detect the five S.F.E.R.A. factors in gamers by developing the “e-S.F.E.R.A. questionnaire” and to measure its statistical properties;
• To explore the relationships of the five e-S.F.E.R.A. factors with more classical psychological dimensions implied in sport psychology, namely, Engagement, Flow at Sport and Gameaholism.

The study was conducted with a convenient sampling method, recruiting respondents by e-mail or online word of mouth, spreading the questionnaire in guilds and groups of different kinds of games. The main Italian esport scenes have been contacted: the Italian Esport Prototype Championship (Machine Simulator Tournament), the PG Nationals (Italian League of Legends Championship), the eSerie A Tim and the Italian Fortnite Championship. The final sample size was 211 people, of which 78.2% were males. Their average age was 26.5 years, and they devoted, on average, 15.32 hours to video games per week. 21.3% participate in official competitions. Shooter games and MMORPGs (massively multiplayer online role-playing games) were the most popular genres (23.8% and 20% of respondents, respectively), followed by i-racing (14.8%), MOBA (multiplayer online battle arena, chosen by 11.4% of respondents) and card games (10%). Finally, the last demographic question was dedicated to the self-perceived level of professionalism: 94 people rated themselves as casual players (44.5%), defined as someone who games just amateurly, for fun; 85 as intermediate (40.3%), occasionally participating in ranked matches; and 32 as advanced (15.2%), someone who participates in high-level competitions.

It is important to underline the discrepancy that can be seen between a self-assessment of respondents’ expertise level (only 15.2% define themselves as an advanced player) and a relative behavioral indicator (as many as 21.3% have participated in official competitions). These results show how difficult it is to clearly define who is and who is not an eplayer [31].

e-S.F.E.R.A: The focal point of the research was the development of the e-S.F.E.R.A. Questionnaire. Items were formulated by two experts (a full professor of applied psychology and an expert sport psychologist certified by the International Olympic Committee). The questionnaire was created and then administered in Italian, following previous questionnaires based on S.F.E.R.A. models related to sports performance [32], business [33] and university teaching areas [34]. Both negatively and positively worded items were formulated to avoid acquiescence or agreement bias [35]. A starting pool of 83 items (18 for Synchrony, 17 for Strengths, 15 for Energy, 16 for Rhythm, and 17 for Activation) was created. During the next step, as suggested by the procedure described by DeVellis (2016), two experts (i.e., sport psychologists who daily use the S.F.E.R.A. model in their psychological practice with athletes) were involved to evaluate if items could properly refer to each dimension they were created for, based on their knowledge and experience. A total of 23 items (6 from Synchrony, 5 from Strengths, 3 from Energy, 4 from Rhythm, and 5 from Activation) were deleted from the starting pool, following experts’ suggestions. A 12-item scale was then reachedfor each factor of the model, to capture their main aspects. Moreover, to evaluate its face validity and ensure the comprehensibility of the item formulation, a pilot administration of the questionnaire was carried out, involving some ex esporters or players of competitive video games without participating in tournaments. Some items’ formulations were adjusted following their suggestions, in order to be more understandable and transversal, i.e. usable by esporters who play different types of games; for example, the item “I moved my avatar with the right timing” was reformulated as “I moved with the right timing”, since not every game has got a single avatar to control. The 60-items version of the instrument was used for the analysis in this work.

The questionnaire requires thinking about a competitive gaming contest that the player recently participated in and responding to a series of statements related to it. The response mode is a Likert scale from 1 (Completely disagree) to 5 (Completely agree).

Flow at Sport: Flow is defined as “a state of consciousness generated while participating in an activity which enjoys and absorbs the individual, and which is intrinsically rewarding” [29]. The I-WOLFS scale evolved from the Work-Related Flow Inventory, originally developed by Bakker [36], which intends to measure Flow within work contexts. Bakker further operationalized this construct into three dimensions: Absorption, an immersive experience where all attentional resources are focused on the current task, even forgetting the passage of time or surrounding events; Work Enjoyment, defined as the subjective pleasure felt during the activity; and Intrinsic Motivation, the desire to execute a task exclusively because of the pleasure directly coming from it [29]. In 2018, Zito and colleagues [29] adapted this scale to the sports context, creating the I-WOLFS scale. Thus, Work Enjoyment has reconceptualized into Sport Enjoyment and Intrinsic Motivation became Intrinsic Sport Motivation. This scale was further modified for this study, to make it adequate to the esportive world, and the items were suitably rewarded. Also in this case, the response mode is a Likert scale, ranging from 0 (Completely disagree) to 6 (Completely agree).

Analyses were performed using both SPSS 28 (IBM, Armonk, NY, USA) and Mplus 8 (Muthén & Muthén, Los Angeles, CA, USA) software. SPSS 28 was implied for: descriptive statistics, reliability (Cronbach’s ɑ), correlations, and regression. Mplus 8 was used to perform CFAs with a Maximum Likelihood method of extraction. CFA (confirmatory factor analysis) was used to test the mono-factorial structure of each S.F.E.R.A. factor; the rationale behind this choice is linked to the practical purpose of this research i.e. to provide sports psychologists with tools to be used separately during psychological sessions aimed at evaluating and optimizing a specific factor. For the sake of completeness, indices of the 5-factor model will also be presented, both for all initial items and those remaining after individual factor analyses.

Common method bias was assessed with Harman’s single factor test [39]. Significant common method variance (CMV) exists if one general factor accounts for the majority of covariance in the measures. Therefore, an exploratory factor analysis (EFA) was performed to test this eventuality. All variables in this study were included in the EFA. Analyses bore out twenty-one factors with eigenvalues greater than one accounting for 62.21% of the total variance, while the first and second factors accounted for 7.28 and 19.29% of the variance, respectively. Considering these results, not a single factor is emerging and accounting for most of the variance, showing the lack of a common method issue.

The factors measured by the e-S.F.E.R.A. Questionnaire showed the following mean points: 3.56 (SD = 0.70) for the Synchrony factor, 3.55 (SD = 0.67) for Strength, 3.59 (SD = 0.63) for Energy, 3.49 (SD = 0.73) for the Rhythm, and 3.26 (SD = 0.90) for the Activation.

Multivariate and univariate skewness and kurtosis of all 60 items were calculated. For multivariate skewness and kurtosis, the tool at the site https:// webpower.psychstat.org/ models/ kurtosis was used (Mardia’s multivariate kurtosis z = 24.36; p < .001); an acceptable value of Mardia’s normalized multivariate kurtosis coefficient is less than 3.0 [40]. Therefore this value shows a non-normal distribution of data. On the other hand, values below ±2.0 for skewness and ±7.0 for kurtosis, according to a more liberal standard [40], are acceptable and the condition of univariate normality. According to these cut-offs, univariate skewness and kurtosis show a normal distribution for all items except for item 4 of Activation (skewness = –2.05).

An EFA (with maximum likelihood method of extraction) was performed for each of the dimensions of the model. In this paragraph will be presented the range of loadings alongside the values of internal consistency (Cronbach’s ɑ) for each dimension of the initial 60-item version of the e-S.F.E.R.A. Questionnaire.

For Synchrony, items explained 21.65% of the variance, with loadings ranging from –.08 to .73 (Cronbach’s ɑ = .69). As regards Strengths, items explained 34.65% of the variance, with loadings ranging from .01 to .78 (ɑ = .81). For Energy, items explained 22.53% of the variance, with loadings ranging from .07 to .73 (Cronbach’s ɑ = .71). For Rhythm, items explained 38.70% of the variance, with loadings ranging from .28 to .81 (Cronbach’s ɑ = .87). Lastly, for Activation, items explained 36.67% of the variance, with loadings ranging from .43 to .79 (Cronbach’s ɑ = .86).

For each single factor, CFA maximum likelihood (ML) estimation was used, except for the Activation factor, where the robust method of estimation (MRL) was used as an alternative [41], considering the non-normal distribution of all items. For the same reasons, the MLR method was used for the five-factor CFAs.

For each CFA of this work, the following cut-offs of the fit indices were followed: values of Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) > .90/.95 indicate a good fit of the model, as well as values of Root Mean Square Error of Approximation (RMSEA) < .05/.08 indicate an acceptable fit (while scores <.10 indicate a mediocre one) and of Standardized Root Mean Square Residual (SRMR) < .08 [42].

The CFA of the starting 12-item scale of Synchrony (M1) showed a not good fit to the data: χ²(54) = 271.792; p <.001; CFI = .60; TLI = .51; RMSEA = .14 (CI:0.12, 0.15); SRMR = .11. Therefore, following the modification indices, alternative models were tried, deleting the following items: Sync_2 (M2); Sync_7 (M3); Sync_8 (M4); Sync_1 (M5); Sync_4 (M6); Sync_12 (M7); Sync_10 (M8). The last model shows the best fit (Table 3). Loadings ranged from .52 to .75 (Fig. 1).

Fig. (1). CFA (maximum likelihood (ML) estimation) standardized solution for the Synchrony factor.

The CFA of the starting 12-item scale of Strength (M1) showed the following not-good-enough fit to the data: χ²(54)= 209.732; p <.001; CFI= .83; TLI= .79; RMSEA= .12 (CI:0.10, 0.13); SRMR= .08. Thus, following the modification indices, alternative models were tried, deleting the following items: Str_12 (M2); Str_3 (M3); Str_4 (M4); Str_5 (M5); Str_6 (M6); and testing the model with the correlation between items Str_9 and Str_7 (M7). The last model shows the best fit (Table 4), with loadings ranging from .36 and .79 (Fig. 2).

Fig. (2). CFA (maximum likelihood (ML) estimation) standardized solution for the Strength factor.

The CFA of the starting 12-item scale of Energy (M1) showed the following not good fit to the data: χ²(54) = 313.924; p <.001; CFI= .57; TLI= .47; RMSEA = .15 (CI:0.13, 0.17); SRMR= .12. Thus, following the modification indices, alternative models were tried, deleting the following items: Ene_9 (M2); Ene_12 (M3); Ene_8 (M4); Ene_2 (M5); Ene_11 (M6); Ene_7 (M7); and testing the model with the correlation between items Ene_4 and Ene_1 (M8). The last model shows the best fit (Table 5), with loadings ranging from .40 and .82 (Fig. 3).

Fig. (3). CFA (maximum likelihood (ML) estimation) standardized solution for the Energy factor.

The CFA of the starting 12-item scale of Rhythm (M1) showed the following not-good-enough fit to the data: χ²(54) = 273.962; p <.001; CFI= .79; TLI= .74; RMSEA = .14 (CI:0.12, 0.15); SRMR= .09. Thus, following the modification indices, alternative models were tried, deleting the following items: Rhy_1 (M2); Rhy_7 (M3); Rhy_9 (M4); Rhy_11 (M5); Rhy_4 (M6); and testing the model with the correlation between items Rhy_3 and Rhy_2 (M7). The last model shows the best fit (Table 6), with loadings ranging from .57 and .82 (Fig. 4).

Fig. (4). CFA (maximum likelihood (ML) estimation) standardized solution for the Rhythm factor.

The CFA of the starting 12-item scale of Activation (M1) showed the following not-good-enough fit to the data: χ²(54) = 430.961; p <.001; CFI= .66; TLI= .59; RMSEA = .18 (CI:0.17, 0.20); SRMR= .12. Thus, following the modification indices, alternative models were tried, deleting the following items: Act_3 (M2); Act_2 (M3); Act_10 (M4); Act_6 (M5); Act_4 (M6); Act_1 (M7); Act_12 (M8); and testing the model with the correlation between items Act_8 and Act_5 (M9). The last model shows the best fit (Table 7), with loadings ranging from .59 and .86 (Fig. 5).

Fig. (5). CFA (maximum likelihood (MLR) estimation) standardized solution for the Activation factor.

Lastly, Table 8 shows the fit indices of the 5-factor models, the first with all 60 items and the second with the remaining 30 items. In the Appendix, we propose an English version of the questionnaire, but future studies need to perform a back translation process to confirm this version [43].

Table 9 highlights strong correlations between the five S.F.E.R.A. factors and all other dimensions, the sole exception is Gameaholism, which seems to correlate with Activation only.

As regard Cronbach’s ɑ, all dimensions show good internal consistency, except the Intrinsic Sport Motivation subdimension of Flow at Sport. Only two out of five items were selected for this dimension, considering that not all items could fit the characteristics of the sample. Specifically, items related to the motivation to play games in free time, which is tautological, and related to the motivation to play linked to retribution (the research was open to both professional and casual players) were not included, choosing instead those that cover the semantic perimeter of the factor.

Regression analysis (Table 10) shows more nuanced relations: only Activation has a strong influence on the dependent variables, including Gameaholism, while Energy has an influence over the Engagement subdimensions of Vigor and Dedication.