Validation of a Measurement Scale on Technostress for University Students in Chile

The main aim in this research was to validate a scale for measuring technostress in Chilean university students under the context of hybrid education. There were 212 university students as participants from the central-south zone of Chile. For measuring technostress manifestations, a technostress questionnaire for Chinese university professors and its adaptation for Spanish university students was used as a base instrument to adapt the scale. The exploratory and confirmatory factor analysis generated an adequacy of the psychometric scale by eliminating three items from the original scales but generated important changes by reordering the other 19 items into only three factors, establishing an important local difference with previous versions that contemplated five factors, but retaining as a central axis the stress produced by a misfit between the person and his or her environment. The resulting scale was based on factors such as Abilities-Demands Techno-Educational, Needs-Supplies Resources, and Person-People Factor. It also has a good internal consistency with a scale that allows for the continuation of technostress measurements in the local context; adding to studies on this topic which have already been carried out on diverse actors of the Chilean educational system; proposing a reliable and valid psychometric scale of technostress in Chilean university students; and giving researchers and academic managers the ability to know the adverse effects of the use of technologies and propose mitigation actions.


Introduction
The incorporation of information and communication technologies (ICT) in education has changed the nature, methods, and processes of learning [1,2]. Nowadays, education has been restructured due to the increasing ICT usage rate by school and university students [3]. According to Talebian et al. [4], ICT enriches existing educational models and provides new technology-based training and learning schemes. At the higher educational level, technological development has facilitated the student exchange between universities and cooperation instances between international students [3]. Likewise, ICTs have challenged and accelerated technological skills development, stimulating "learning by doing" and contributing to sustainable development through the fourth sustainable development goal on quality education [5][6][7].
ICT use, however, can cause stress, which corresponds to a physical and emotional response to distress caused by an individual's imbalance between perceived demands and perceived resources, and their abilities to cope with those demands [8]. When stress is associated with ICT use, it is called technostress, a concept first coined by the American psychiatrist Craig Brod [9], who defined it as a "modern adaptative illness caused by the inability to cope with new computer technologies in a healthy manner" [10] (p. 242). In the last two decades, it has become a topic of growing interest for studies [11,12].
Salanova defines technostress as "a negative psychological state related to the ICT use or threat of its use in the future, conditioned by the mismatch perception between the demands and resources related to the ICT use that leads to a high unpleasant psychophysiological activation level and to the negative attitude development towards ICT" [13] (p. 423). For its part, according to Tarafdar et al. [14], technostress is a product of the inability to adapt or cope with new technologies, and constitutes a process characterized by the presence of technological environmental conditions that are evaluated as demands or techno-stressors by the individual, which set in motion coping responses leading to psychological, physical, and behavioral manifestations.
The new challenges resulting from the COVID-19 pandemic for higher education through ICT [31][32][33][34][35] and given the eruption of hybrid higher education [36,37], an interest in learning about its psychological and mental health effects on students is evident. Given this scarcity of studies on technostress in university students, it is necessary to have psychometric measurement scales adapted and validated to various local contexts to facilitate further study. In this regard, this article aimed to adapt and validate measurement of technostress scale for university students (TS4US) based on previous work by Wang et al. [21] in China, and Penado-Abilleira et al. [38] in Spain. In theoretical terms, this local validation contributes to strengthen the global empirical case set that allows to specify the theoretical constructs incorporated in a psychometric instrument capable of measuring technostress in university students.
In the Chilean case, technostress studies in education have been published in mainstream journals focusing on other members of educational communities, such as teachers [18,19] and managers [25]. In practical terms, this work will allow access to an instrument with structural validity to advance the study in university students.

Materials and Methods
The adaptation for Chilean students of the Wang and Li [21] and Penado-Abilleira et al. [23] technostress questionnaires began with a re-translation of the items that compose the original Wang and Li [21] scale to the Spanish of current use in Chile by a specialized linguist who was part of the research team, and adapting the items for a university student population as well as establishing local semantic differences in comparison with Penado-Abilleira et al. [23]. The result was pre-tested with a group of 20 students to evaluate their comprehension. The resulting scale is presented in Appendix A (in Chilean Spanish), keeping a 5-point Likert scale: Strongly Disagree = 1, Disagree = 2, Neither Disagree nor Agree = 3, Agree = 4, and Strongly Agree = 5. The questionnaire has been self-administered online, after giving informed consent response, collecting the survey without respondent identification, and only presenting non-potentially identifiable human data.
Then, SPSS 23 software was used (IBM, New York, NY, USA) to analyze the 22-item TS questionnaire [21,38], for establishing psychometric validity by means of structural evidence [39]. As a first step, a univariate descriptive statistical analysis was applied, with emphasis on variance (>0), skewness and kurtosis (|≤1|, both). To measure confidence levels, the authors applied the Kaiser-Meyer-Olkin measure of sampling adequacy (KMO). Moreover, the authors used Bartlett's test of sphericity to identify items belonging to the factors within the scale as a form of exploratory factor analysis (EFA) with the extraction method, unweighted least squares (ULS), rotation method, and Oblimin with Kaiser normalization [40]. Then the authors analyzed the exploratory factors by using a confirmatory factor analysis (CFA) with FACTOR software [41]. In addition, they used the Hull method to select the number of factors according to the EFA results, including high communalities, high factor loadings to support sample size, and minimum items per factor (MIF) [42][43][44], considering the comparison possibilities with Wang et al. [21] and Penado-Abilleira et al. [38]. It was necessary to obtain a report of the indicators: Chisquare/degree freedom ratio (χ2/df), root mean square error of approximation (RMSEA), adjusted goodness-of-fit index (AGFI), goodness-of-fit index GFI, comparative fit index (CFI), (RFI), normed fit index (NFI), non-normed fit index (NNFI), and Root Mean Square of Residuals (RMSR) [45]. See Table 1. ≥0.05 ≤0.08 ++ NR: not reported. ** Good fit; * acceptable fit. + indicated in Penado-Abilleira et al. [38]. ++ indicated in Kalkan et al. [47].
Finally, the internal reliability of the resulting instrument will be validated by calculating Cronbach's Alpha by SPSS 23 software [48,49].

Sample Characterization
The Technostress Scale for University Students (TS4US) was applied in the first academic semester 2021 to a set of 212 participants (≥ 200, overcoming small sample sizes for factorial analysis) [43], university students from the Chile Central-South Zone, which concentrates more than 70 percent of the national and university population [50,51], as shown in Table 2 (Available data in Table S1: TS4US_data_22_var, Supplementary Materials).

Exploratory Factor Analysis
Firstly, we analyzed the possible prevalence of the factors identified by Wang and Li [21] for Chinese university teachers and Penado-Abilleira et al. [38] for Spanish university students. These five original factors establish the stresses for technology use between personal capabilities and organizational demands (abilities-demands organization, ADO), personal capabilities and technology demands (abilities-demands technology, ADT), personal needs and organizational resources to perform their tasks (needs-supplies organization, NSO), personal needs and their own available technology resources (needs-supplies technology, NST), and interpersonal relationships between students (person-people factor, PPF). Univariate descriptive statistical analysis was applied, and no ordinal variable presented variance equal to zero, so all of them contributed to the common variance. But in terms of skewness and kurtosis, 3 variables present kurtosis (kurt) problems: VAR_05 (kurt = −1.109), VAR_10 (kurt = −1.281), and VAR 22 (kurt = −1.160), see Figure 1.

Exploratory Factor Analysis
Firstly, we analyzed the possible prevalence of the factors identified by Wang and Li [21] for Chinese university teachers and Penado-Abilleira et al. [38] for Spanish university students. These five original factors establish the stresses for technology use between personal capabilities and organizational demands (abilities-demands organization, ADO), personal capabilities and technology demands (abilities-demands technology, ADT), personal needs and organizational resources to perform their tasks (needs-supplies organization, NSO), personal needs and their own available technology resources (needssupplies technology, NST), and interpersonal relationships between students (personpeople factor, PPF). Univariate descriptive statistical analysis was applied, and no ordinal variable presented variance equal to zero, so all of them contributed to the common variance. But in terms of skewness and kurtosis, 3 variables present kurtosis (kurt) problems: VAR_05 (kurt = −1.109), VAR_10 (kurt = −1.281), and VAR 22 (kurt = −1.160), see Figure 1.  Tables 3 and 4 show the unrestricted results of the exploratory factor analysis preserving the 18 variables and determining with SPSS 23 a KMO of 0.897 and Bartlett's test with a Chi-square of 2432.170 with 171 degrees of freedom and a significance level of 0.000 for the four factors TS4US instrument. The authors achieved a 59.476% explained variance proportion. Although these results were individually valued as positive, it was also observed that factor 4 did not comply with the minimum variables recommended per factor (>3) [42][43][44], which led us to test another variable reduction alternative.    Tables 3 and 4 show the unrestricted results of the exploratory factor analysis preserving the 18 variables and determining with SPSS 23 a KMO of 0.897 and Bartlett's test with a Chi-square of 2432.170 with 171 degrees of freedom and a significance level of 0.000 for the four factors TS4US instrument. The authors achieved a 59.476% explained variance proportion. Although these results were individually valued as positive, it was also observed that factor 4 did not comply with the minimum variables recommended per factor (>3) [42][43][44], which led us to test another variable reduction alternative.

Confirmatory Factor Analysis
Additionally, the authors successfully adapted the 19 variables analyzed dataset for 19 variables to confirmatory factor analysis (CFA) with the use of the FACTOR software. The CFA obtained a KMO-Kaiser-Meyer-Olkin-equal to 0.89743 (>0.8) and Bartlett's test of sphericity with a Chi-Square 9668.9 with 171 degrees of freedom and a significance level of 0.000010. Those results were significant and good enough to present the adequacy of the Pearson correlation matrix.
The Hull method for selecting the number of three factors was implemented with an adequacy of the Pearson correlation matrix. Then the authors reduced the TS4US questionnaire according to its latent variables in three factors (see Table 7).  Table 8 sets out the proposed model results in comparison with the resulting validity and reliability values in Wang et al. [21] and Penado-Abilleira et al. [38], for the RMSEA, AGFI, GFI, CFI and RMSR indicators by FACTOR software. In comparative terms, the proposed model reports an RMSEA with an acceptable fit equal to Wang et al. [21], AGFI and GFI with a good fit equal to Penado-Abilleira et al. [38], CFI with a good fit in contrast to the acceptable fit reported by Wang et al. [21], and RMSR with a good fit in contrast to the acceptable fit reported by Penado-Abilleira et al. [38]. ≥0.05 ≤0.08 ++ NR: not reported. ** Good fit; * acceptable fit. + indicated in Penado-Abilleira et al. [38]. ++ indicated in Kalkan et al. [47].
Finally, Table 9 shows the instrument's internal reliability by SPSS 23 software, with a total Cronbach's Alpha of 92.5% for the set of 19 items, whose definitive scale is presented in Chilean Spanish in Appendix B.

Discussion
In this research, through an exploratory and confirmatory factor analysis, the factors identified by Wang and Li [21] for Chinese university teachers and Penado-Abilleira et al. [38] for Spanish university students were analyzed in the context of Chilean university students from public and private institutions. As a result of these analyses, within the Chilean university students participating in this study, the model explaining technostress was reduced from five to three factors, with the loss of three variables 5 (ADO05), 10 (ADT01), and 22 (PPF04), a product of the instrument's adjustment to the specific sample composition; maintaining factors (theoretical constructs) that made it possible to measure the phenomenon of technostress. It was observed that the variables that make up the measurement scale tend to be grouped into the main factors that are part of various theories that explain stress, as detailed below.
The variables corresponding to the tensions between personal needs and organizational resources (NSO) and personal needs and technological resources (NST) are grouped in the first factor; it is known that the availability and usability of resources is a moderating factor on stress [52]. It should be noted that, in this factor, the variables that make up the personal needs or demands and technological resources (NST) that are associated with the self-perception of usefulness of available ICTs are grouped together, resulting in a factor on personal needs and available resources (needs-supplies resources, NSR). In this regard, the perceived usefulness of ICT use has been defined as a factor which inhibits technostress [53,54].
Then, there is a second factor, which groups those variables associated with interpersonal relationships (PPF) among students, maintaining this factor according to Wang and Li [21] and Penado-Abilleira et al. [38], which is associated with the social support given by peers and peer learning in the face of novel and potentially distressing processes [2,12], which constitutes a protective factor mitigating technostress [55].
Also, the third factor includes those variables associated with the relationship between personal capabilities and organizational demands (ADO), and personal capabilities and technological demands (ADT). If work demand is not in line with capabilities, this can lead to stress manifestations [56][57][58][59]. On the other hand, it includes variables on personal needs and technological resources (NST), but those associated with technological overload, which might be a cause for technostress rather than from the organizational type [9,60,61]. Because of high demand and lack of resources for working, ICT is associated with increased technostress [13,52]. In sum, the third factor relates personal capabilities and techno-organizational demands, specifically techno-educational demands (abilitiesdemands techno-educational, ADTE).
Finally, in the absence of a meta-analysis to ensure a broad application of a valid scale [62,63], the local validation is a contribution to the global empirical case set to clarify the theoretical constructs incorporated in a psychometric instrument with the ability to measure technostress in university students in different countries and cultures, as well as its practical implications in local terms to expand the technostress studies in university students.

Conclusions
This article validated a scale to measure technostress in Chilean university students in the context of hybrid education, based on activities combined and carried out in different environments (physical or virtual) and times (synchronous or asynchronous) [64,65], using as a basis a technostress questionnaire for Chinese university teachers and its adaptation for Spanish university students. The exploratory and confirmatory factor analyses eliminated three items from the original scales but generated important changes by reordering the other 19 items in only three factors, establishing an important difference with the previous versions that contemplated five factors, with a good internal consistency and having as a central axis the stress produced by the misfit between the person and his/her environment. Thus, the instrument allowed for the creations of a scale for measuring technostress in Chilean university students (TS4US) based on the following factors: Abilities-Demands Techno-Educational (ADTE), Needs-Supplies Resources (NSR), and Person-People Factor (PPF). It will also allow for a continuation of the measurements of technostress in a local context, increasing the studies on this topic already carried out in different actors of the Chilean educational system.
Although all the parameters that ensure the quality of the sample size were met in this study and the sample exceeded the small sample cut-off for a factor analysis , a limitation of this work was not having a larger sample, which we will address after the validation of this psychometric instrument. On the other hand, the article has been limited only to the validity analysis on the psychometric measurement scale studied, without analyzing other dimensions. Also, future research lines will allow for a series of local studies on technostress in the Chilean educational system [18,19,25,67] to be completed, as well as for extensive measurements to be developed on technostress in university students, and further work relating technostress with other aspects, such as techno-addiction, cyberbullying, transformation of learning processes and engagement to educational work, and stimulating 'learning by doing' [5,6,[68][69][70]. Funding: The publication fee (article processing charge, APC) was partially funded by Universidad Central de Chile (Code: ACD219201) and was partially funded through the publication incentive fund by Universidad Andres Bello (Code: CC21500), and the Universidad Autónoma de Chile (Code: CC456001).
Institutional Review Board Statement: Not applicable, no animal and human studies are presented in this manuscript. And no potentially identifiable human images or data is presented in this study.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study, by initial question in a self-administered survey.

Conflicts of Interest:
The authors declare no conflict of interest.