Adaptation and psychometric properties of the SBI-U scale for Academic Burnout in university students

Título: Adaptación y propiedades psicométricas de la escala SBI-U sobre Burnout Académico en estudiantes universitarios. Resumen: El objetivo del presente trabajo ha sido desarrollar una adaptación española a estudiantes universitarios de la escala School Burnout Inventory (SBI) de 9 ítems. Así, se llevó a cabo una doble adaptación, por un lado, del inglés al español y, por otro, de estudiantes de secundaria a universitarios. La escala se aplicó a 578 estudiantes universitarios (25.7% hombres; 74.3% mujeres) de diversas zonas de España. Los resultados muestran que el SBI-Universitarios tiene la misma estructura que la versión original inglesa para estudiantes de secundaria, ello se ha constatado el mediante análisis factorial confirmatorio que ha indicado la existencia de tres factores: Agotamiento, Cinismo e Inadecuación. Por otro lado, las tres subescalas presentan una fiabilidad aceptable (entre .77 y .70). Además, se aportan indicios de validez a través de dieciocho correlatos externos y siete escalas de contraste. Finalmente, el SBI-U se configura como un instrumento potencialmente útil para evaluar el burnout académico en estudiantes universitarios. Palabras clave: Burnout académico; SBI; universidad; versión española; estudio instrumental. Abstract: The objective of the present study was to draw up a Spanish adaptation for university students of the School Burnout Inventory (SBI) 9item scale. This entailed a double adaptation, on the one hand from English into Spanish and then from secondary school students to university students. The scale was applied to 578 university students (25.7% men; 74.3% women) from different regions in Spain. The findings indicate that the University students-SBI has the same structure as the original version in English for secondary school students. This was confirmed by factor analysis that pointed to the existence of three factors: Exhaustion, Cynicism and Inadequacy. Furthermore, the three subscales showed acceptable reliability (between .77 and .70) In addition to this, indications of validity were found using eighteen external correlates and seven contrast scales. Finally the SBI-U constitutes a potentially useful instrument for evaluating academic burnout in university students.


Introduction
In the mid seventies of the last century, Freudenberger used the construct burnout to define the deterioration process in the care and professional attention that users receive in services organisations such as healthcare and education centres, etc.Later on, Maslach (1976) referred to the burnout experienced by employees who provide services to people.Maslach and Jackson (1981) defined it as an inadequate response to chronic emotional stress, whose main features are physical, psychological and emotional exhaustion (emotional fatigue), a cold attitude towards others (depersonalization) and a feeling of inadequacy concerning the tasks they have to carry out (low personal fulfilment).
However, as some authors point out, (Caballero, Hederich, & Palacio, 2010;Salmela-Aro, Kiuru, Leskinen, & Nurmi, 2009) we need to construct and validate specific scales to evaluate academic burnout in particular.Hence, the contribution of this study is that it presents an instrument that enables us to evaluate academic burnout in university students given that a scale of these characteristics has never been published before in Spanish.
Following on from the above the present instrumental study (Montero & León, 2007) sets three objectives.Based on studies of burnout in job environments (Schaufeli, Bakker, & Salanova, 2006) and in academic environments (Salmela-Aro et al., 2009;Salmela-Aro & Kunttu, 2010;Martínez et al., 2002, Schaufeli et al., 2002) it puts forward an academic burnout model for university students.In the first objective 1, the Salmela-Aro et al., (2009) academic model in Spanish is expected to have three factors: Exhaustion due to the demands of university work, Cynicism concerning the meaning of university, and A sense of inadequacy at university.As far as reliability is concerned, the Cronbach alphas are expected to be acceptable (objective 2) for all three factors, and finally, the convergent validity (objective 3) was checked by correlating the three academic burnout factors with various indicators such as social demographic data, irritation, psychological wellbeing, depression, personality, self-regulation and selfefficacy.

Method Participants
The sample was made up of 578 university and postgraduate students from different Autonomous Communities in Spain.The data was gathered by three Spanish universities (Universitat Rovira i Virgili -Tarragona, Universidad de Salamanca and Universidad de Valladolid) and one EADA Business School (Escuela de Alta Dirección y Administración).25.7% of the sample were men and 74.3% women (M= 21.6;SD=4.94).The types of studies were undergraduate (61.9%), diploma (28%), graduate studies (9%) and masters (1%).49.9% of the sample were in their first year, 20% in the second, 9.2% in the third, 15.4% in the fourth, 25% in the fifth and 15% were on postgraduate or masters courses.

Instruments
-The School Burnout Inventory (SBI-9;Salmela-Aro et al., 2009) (See Annex 1) was drawn up for the purpose of evaluating burnout among adolescents in secondary education (16 year olds).The version we present has undergone a double adaptation.It was first translated from English into Spanish and the contents were then adapted from secondary school students to university students.The first adaptation was carried out following the steps set down in the scientific literature on adapting evaluation instruments (Muñiz & Bartram, 2007): Translation by experts of the items into Spanish, a discussion group on the translation of the items followed by their back-translation into English and a final check to make sure that the two versions are equivalent.Group techniques such as brainstorming and focus group were used in the second adaptation and were applied to various groups of university students.The scale consists of 9 items assessed on a Likert type scale (from 1=Completely disagree to 6=Completely agree) and 3 subscales (Exhaustion, Cynicism and Inadequacy).In order to distinguish it from the original version the Spanish version was labelled SBI-U.-The Spanish version of the Irritation Scale (IS; Merino, Carbonero, Moreno, & Morante, 2006) consists of 8 items and 2 subscales.It enables us to diagnose irritation, which is defined as a state of progressive psychological fatigue that cannot be relieved with normal rest periods.Irritation may also arise when a person feels there is a discrepancy between a given state of affairs and their achievement of an important personal goal.The first subscale is called Emotional irritation (reliability = .86)and is made up of 5 items (for example, "3.-When other people talk to me it makes me irritated"); the second is called Cognitive irritation (reliability = .87)and is made up of 3 items (for example, "1.-I find it hard to switch off after work").The Likert responses were gathered using a 7 point scale (from 1.-Very much disagree to 6.-Very much agree).).-The Self-Regulation Scale (SRS; Luszczynska, Diehl, Gutiérrez-Doña, Kuusinen, & Schwarzer, 2004) consists of 7 items which are answered using a four point scale (1=not at all to 4=a lot) and its configuration is one dimensional.Self-regulation is regarded as a dispositional variable that is responsible for being able to regulate oneself in a wide range of situations.Hence controlling alertness is a key component when it comes to overcoming obstacles.The scale's internal consistency in the different transcultural studies that have been carried out ranges between .63 and .87.Here is one example of an item: "7.-I keep focussed on my goals and do not allow anything to divert me from them".-The General Self-Efficacy Scale (GSES; Sanjuán, Pérez, & Bermúdez, 2000) assesses the stable feeling of personal competency in coping effectively with a large variety of stressing situations.This is a one-dimensional instrument made up of 10 items with a 10-point response format (1= Completely disagree to 10= Very much agree).An example of one item is: "5.-Thanks to my qualities and resources I can overcome unforeseen situations".The studies were carried out with samples featuring different nationalities and showed considerable internal consistency for the scale (between .79 and .93).-Finally, we also used certain correlates, also known as external indicators (Streiner & Norman, 2008), in the form of questions that the respondents had to answer.One of them was answered using a scale of 10 anchors (1=Completely disagree to 10=Very much agree).These referred to support from tutors and professors, the class schedule load, the furniture and classroom conditions, and professor non-attendance of classes without prior notice.Others were answered providing specific details or stating how often they did certain things.For example, students were asked how often they had woken up at night thinking about things related to a certain subject, what they had studied over the weekends, etc.

Procedure
Non-probabilistic sampling was used, which is also known as random accidental sampling.Students participated on a voluntary and anonymous basis and their personal identification data was not registered.The data was always gathered inside the classroom on a group basis following the consent of the professor teaching the subject.An expert psychologist was always present during the application of the questionnaire to solve any possible queries that the university students might have.

Data Analysis
Taking into account that we already had a hypothesis concerning the questionnaire's factor structure, given that the original version had put forward three factors, we carried out a confirmatory factor analysis (CFA) in order to verify this structure.Once the scale's factor structure was defined we used other scales and external indicators to analyse its reliability and criterion validity.The data were analysed using the programmes SPSS 19.0 and Mplus 5.1.

Results
A CFA was performed on the polychoric correlation matrix given that the items seemed to be on an ordinal scale.We used the fit indices put forward by Hu and Bentler (1999), the comparative fit index (CFI), the non-normative fit index (NFI) and the root mean square error of approximation (RMSEA).Although there is no absolute consensus, acceptable cut-off values for the first two tend to be equal to or above .90whereas values below .08 were considered acceptable and equal to or under .05excellent for the RMSEA (Browne & Cudek, 1993).
In our case the SBI-U showed an acceptable fit with the three factor model initially put forward by Salmela-Aro et al. (2009), and it came up with the following indicators: CFI = .95;TLI = 0.93; RMSEA = .06(90% reliability interval, .047-.078; RMSEA approximate fit test <.05, p=.098).Consequently the indicator values were close to optimum.
Figure 1 shows the resulting model.Although we had initially considered a three factor structure, given the close relationship between two of them, we thought that a two factor model might be more suitable than the three factor structure.However, when the model was contrasted using CFA, creating a new factor by joining the items of factors 2 and 3, the model's fit decreased significantly (  = 26.72;p <.001).
anales de psicología, 2015, vol.31, nº 1 (enero) The questionnaire's dimensionality was established and Table 1 shows the scales' descriptive statistics, reliabilities and validity ratios.Reliability is between .70 and .77which means that it can be considered acceptable if we take into account the small number of items that make up each of the subscales.
Table 1 also displays the correlations between the three scales and eighteen external correlates and five scales (Irritation, BDI, OPERAS, SRS and GSES).The three academic burnout factors (Exhaustion due to the demands of university work, Cynicism concerning the meaning of university and A sense of inadequacy at university) displayed relations with the different measures in the expected direction and correlated positively with irritation (emotional and cognitive) and general depression, whereas the third factor also showed a positive correlation with emotional stability (Sense of inadequacy at university).Fur-thermore, the three factors displayed negative correlations with extraversion, self-regulation and general efficacy.More specifically, we found different negative correlations such as between: Exhaustion concerning university work (F1) and emotional stability; Cynicism concerning the meaning of university (F2) and emotional stability, responsibility and amiability; and between A sense of inadequacy at university (F3) and responsibility, amiability and openness to experience).
As regards the external correlates, the three subscales that make up academic burnout were associated with various of these.We found three direct correlations, (for example, the number of the academic year, the Sundays they had devoted to studying, competition among colleagues, faculty absenteeism, etc.) plus nineteen inverse correlations (such as for example, support from tutors and professors, the class schedule load, level of comfort of furniture, transport, etc.).
Table 1.SBI-U-9: Descriptive statistics, reliability, confidence intervals, correlations among the three subscales, and the correlations between the three subscales and social-demographic aspects, external correlates and the contrast scales..00-.17** -.15**I have an acceptable schedule load.
-.13** -.11** -.12**I receive adequate feedback from my classmates when I do projects and/or prepare for exams.
-.05 -.13** -.15**I receive adequate feedback from my professors when I do projects and/or prepare for exams.

Discussion
In the present study we present the psychometric properties of the SBI-U-9, a brief instrument that enables us to evaluate academic burnout (exhaustion, cynicism, inadequacy) in university students and this is the first time this scale has been presented in Spanish.
The second objective was corroborated given that the reliability of the three factors was found to be appropriate.Salmela-Aro et al. (2009) reliability ratios for the three subscales were: Exhaustion (.80), Cynicism (.80) and Inadequacy (.67).The ratios in the Spanish version were acceptable ranging between .70 and .77.Moyano and Riaño-Hernández's (2013), adaptation into Spanish the SBI, and tested on adolescents obtained similar reliabilities: Exhaustion (.57), Cynicism (.63) and Inadequacy (.49).As regards the MBI-SS (Schaufeli et al., 2002) the resulting reliabilities in university students for the three subscales that make up the scale were as follows: From .74 to .79 (Spanish), from .69 to .82(Portuguese) and from .67 to .86 (Dutch).
As far as criterion validity is concerned, a review of the literature shows that burnout is related to irritation (Merino-Tejedor et al., 2006), depression (Ahola & Hakanen, 2007), personality (Hudek-Knezevic, 2011), self-regulation (Gurcay, 2009), self efficacy (Skaalvik, 2007) and certain external indicators (Salanova et al., 2000) such as age, profession, etc.The results of the present study found significant correlaanales de psicología, 2015, vol. 31, nº 1 (enero) tions between the three factors of the SBI-U scale and the abovementioned variables.They thus meet our objective in that they confirm the proposed validity ratios.In the study we are presenting what stands out primarily is the correlation between academic burnout and depression, which ranges between .29 and .41.Although the instruments used to measure depression were different, the association between academic burnout and depression was also found by Moyano and Riaño-Hernández (2013) in their research study, where the correlations ranged from .31 to .48.
By way of conclusion we can state that the SBI-U-9 scale's psychometric properties, in its Spanish version, are satisfactory.Thus, the CFA clearly confirms the structure of the three original factors.Furthermore, the reliability estimates were acceptable and there is clear evidence of validity.We can consequently state that the present scale in Spanish is useful for evaluating academic burnout in university students.
The study has at least two drawbacks: On the one hand, more longitudinal studies need to be carried out in order to further determine the relationship between academic burn-out and other variables such as the year of the student's course (for example, the first year versus the fourth year), the number of repeat subjects, the type of repeat subjects (for example, education, psychology, medicine, engineering, economics, law, etc.), etc.Furthermore, the validation of a psychometric instrument is a dynamic process which does not end once it has been constructed and published (Padilla, Gómez, Hidalgo, & Muñiz, 2006).Hence new studies are likely to contribute new data both on academic burnout and on the scale we have presented.Finally, this study uses a convenience sample to adapt the scales and this limits the validity of the findings.Having said this, this is relatively common practice in research studies in the behavioural sciences (Alonso-Tapia & Rodríguez-Rey, 2012;Boada-Grau & Gil-Ripoll, 2011;Boada-Grau, Sánchez-García, Prizmic-Kuzmica, & Vigil-Colet, 2012).However, as Highhouse and Gillespie (2008) point out, quite often the use of incidental samples does not constitute an important threat to a study's validity and on the other hand, the present study, despite the fact that the sample is a convenience sample, incorporates students from different faculties.