British Journal of Educational Technology doi:10.1111/bjet.12832 Vol 0 No 0 2019 1–15 “I am fine with any technology, as long as it doesn’t make trouble, so that I can concentrate on my study”: A case study of university students’ attitude strength related to educational technology acceptance Nicolae Nistor*, Dorin Stanciu*, Thomas Lerche* and Ewald Kiel Nicolae Nistor is a senior research fellow at the Faculty of Psychology and Educational Sciences, LudwigMaximilians-Universität München, Germany, and senior contributing faculty at Richard W. Riley College of Education and Leadership, Walden University, USA. His research focus includes teaching and learning with technology, educational technology acceptance, knowledge building communities, and learning analytics. He is currently an associate editor of Computers in Human Behavior. Dorin Stanciu is an associate professor at the Department of Psychology and Education of the Technical University of Cluj-Napoca, and contributing faculty at Richard W. Riley College of Education and Leadership, Walden University, USA. He is a senior researcher in User Experience and Technology Acceptance, Ambient Assisted Living and Human Robot Interaction in several past and present European research programs. He is also interested how augmented reality and the Internet of Things can enhance human learning. Thomas Lerche is a senior lecturer of the School Education at the Department of Education and Rehabilitation at Ludwig-Maximilians-Universität Munich, Germany. His main research interests are performance assessment, competence-oriented testing with multiple choice tests and activation of learning processes. Ewald Kiel is Chair for School Education in the Department of Education and Rehabilitation, and a senate member of the Ludwig-Maximilians-Universität Munich, Germany. His main research interests are teacher professionalization, inclusion, school development and intercultural education. Address for correspondence: Nicolae Nistor, Faculty of Psychology and Educational Sciences, Ludwig-Maximilians-Universität, Leopoldstr. 13, D-80802 München, Germany. Email: nic.nistor@uni-muenchen.de Abstract Technology acceptance models presuppose that technology users have clearly defined attitudes toward technology, which is not necessarily true. Complementary, socialpsychological research proposes attitude strength (AS), a construct that has been so far insufficiently examined in the context of technology acceptance. Attitudes toward technology might become weaker after frequent changes in the used technology. This study examines the relationships between AS and educational technology acceptance predictors. In the case of N = 225 German undergraduate students of Educational Sciences, “millennials” using the learning management system Moodle, and based on structural equations modeling (SEM) and fuzzy set qualitative comparative analysis (fsQCA), we found significant relationships between AS and acceptance predictors. Further results suggest two situations leading to technology acceptance, one in which students are performance-oriented and comply with faculty recommendations; the other in which students are technically experienced and will accept any technology, but avoid technical problems and effort. While the latter situation is only vaguely suggested by SEM, it is much clearly indicated by fsQCA. For acceptance research, we conclude that current acceptance models should be extended by AS, and employ fsQCA. For educational practice, we recommend using fsQCA to assess acceptance predictors when educational technology is implemented in higher education. *The first three authors have equal contributions to the paper. © 2019 British Educational Research Association 2 British Journal of Educational Technology Vol 0 No 0 2019 Practitioner Notes What is already known about this topic • Educational technology acceptance is mainly represented by the use intention of that technology, further predicted by attitudes operationalized as performance and effort expectancy and social influence. • Attitude strength (AS) can differ interindividually, and be directly related to attitudes or moderate the relationships between them. • Little is known about the relationships between AS and technology acceptance models. • Technology acceptance models are currently verified by regression and structural equations (SEM); fuzzy sets qualitative comparative analysis (fsQCA) additionally informs about factor configurations leading to an outcome. What this paper adds • In a sample of over 200 German undergraduate students of Educational Sciences using Moodle as a learning management system, UTAUT could be partially verified. • In one situation leading to technology acceptance, students are performance-oriented and comply with faculty recommendations. • In another situation leading to acceptance, students are technically experienced and will accept any technology, but avoid technical problems and effort. • While the latter situation is only vaguely suggested by SEM, it is much clearly indicated by fsQCA Implications for practice and/or policy • AS may interindividually differ in response to frequent technology changes, and play a significant role for educational technology acceptance. • Technology acceptance models used to assess the implementation of educational technologies in higher education can be extended by fsQCA, in order to identify configurations of acceptance factors leading to technology acceptance. • Elaborating on different patterns, or situations, of use can lead to targeted educational interventions that enhance acceptance behaviors and, in the case of educational technology, that enhance learning. Problem statement Technology acceptance models, such as the Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh, Morris, Davis, & Davis, 2003; Venkatesh, Thong, & Xu, 2012) presuppose that technology users have clearly defined attitudes toward technology, which is not necessarily true. Complementary, social psychological research (Eaton, Majka, & Visser, 2008; Howe & Krosnick, 2017; Visser, Bizer, & Krosnick, 2006) proposes the construct of attitude strength (AS), first defined by Petty and Cacioppo (1986) as the degree to which attitude manifests itself in the form of temporal persistence, resistance to counter persuasion and predictability of behavior. AS may play a substantial role in the mechanisms of technology acceptance. Moreover, AS might be influenced by frequent changes in the technology they use, from minor updates to radical changes of platforms, which is currently frequent (Orlando, 2014). However, AS has been so far insufficiently examined in the context of technology acceptance (Bhattacherjee & Sanford, 2009; Kim, Chun, & Song, 2009), although each change, and especially major changes rise © 2019 British Educational Research Association Attitude strength and educational technology acceptance 3 the question whether and to what extent the users will accept the new technology (eg, Scherer, Siddiq, & Teo, 2015). In this study, we examined the case of German university students, “millennials” (Hur, Lee, & Cho, 2017) using the learning management system (LMS) Moodle after frequent changes and updates, and in parallel with other LMS and further software. Against this background, we aimed to understand the role AS may play in connection with the UTAUT (Venkatesh et al., 2003, 2012). To overcome the limitations, traditional methods such as regression and structural equations modeling (SEM) imposed to data analysis, we employ fuzzy sets qualitative comparative analysis (fsQCA) (Ragin, 2008) to find configurations of variables that lead to educational technology acceptance (Martí-Parreño, Galbis-Córdova, & Miquel-Romero, 2018). The remainder of this paper presents the theoretical and methodological framework of the study, the findings and their interpretation and draws conclusions pertaining to educational technology acceptance research and practice. Theoretical framework The TAM family Current technology acceptance models rely on a double definition of technology acceptance, as behavioral intention and as actual technology use (Davis, 1985). The models are rooted in the social psychology, applying the Theory of Reasoned Action (TRA) and its extended version, the Theory of Planned Behavior (TPB) that postulate a link between intention and behavior, with intentions being predicted by attitudes (Ajzen & Fishbein, 2000). At least three technology acceptance models emerged over time from TRA/TPB, the TAM (Davis, 1985), TAM2 (Venkatesh & Davis, 2000) and TAM3 (Venkatesh & Bala, 2008). From the many acceptance predictors examined in these models, Venkatesh et al. (2003) identified performance expectancy (PE), effort expectancy (EE), social influence (SI) and facilitating conditions (FC) as main acceptance factors, with PE, EE and SI predicting the behavioral intention, and FC directly impacting the technology use behavior. This Unified Model of Acceptance and Use of Technology (UTAUT) was revalidated by Venkatesh et al. (2012), adding a few new predictors to PE, EE, SI and FC. From several conceptual and methodological limitations that the UTAUT displays (see, for instance Benbasat & Barki, 2007; Nistor, 2014), one that is particularly relevant for this study is the assumption that technology users exhibit clearly defined attitudes that impact their behavior, and thus induce technology adoption as a rational decision (Schoonenboom, 2012). While this may be true in some cases, it is only half the story. People may use technology enthusiastically, uncritically and regard it as fashion (Wang, 2010), they may resist technology adoption (Sanford & Oh, 2010), or they may be indifferent to which technology they use—we dicuss the latter possibility below. A notable methodological limitation is that the TAM family models were sofar exclusively validated and applied using regression and SEM (Al-Emran, Mezhuyev, & Kamaludin, 2018; Khechine, Lakhal, & Ndjambou, 2016; Šumak, Heričko, & Pušnik, 2011), which takes into consideration only the presence of predictors. However, various studies (eg, in the special issue on educational technology acceptance edited by Nistor, 2014) show that technology users can develop use intention and actually use the technology in cases when one or more acceptance predictors are absent, as well. For instance, in different cases of educational technology implementation, Pynoo and van Braak (2014) found EE as the only significant predictor of use intention, while AgudoPeregrina, Hernández-García, and Pascual-Miguel (2014) report on the converse situation, in which there was no significant relationship between EE and use intention. Pappas, Giannakos, and Sampson (2019) propose a different method, the fuzzy set qualitative comparative analysis © 2019 British Educational Research Association 4 British Journal of Educational Technology Vol 0 No 0 2019 (fsQCA) that, unlike SEM, reveals configurations—ie, the presence and absence of predictors— that lead to acceptance and adoption of a technology (Ragin, 2008). Pappas et al. (2019) suggest that fsQCA can deepen our understanding of technology acceptance. Additional research is needed to unfold the potential of this method. While Martí-Parreño et al. (2018) applied fsQCA to explain students’ attitudes toward educational video games, we are not aware of any fsQCA study of current acceptance models such the UTAUT, and including attitude strength. Attitude strength Attempting to develop a more differentiated view of the attitude-behavior link postulated by the TRA/TPB (Ajzen & Fishbein, 2000), the social-psychological research of the last decades introduced the concept of attitude strength (Howe & Krosnick, 2017). Accordingly, the question is no longer whether, but to what extent attitudes influence behavior. Thus, attitude strength (AS) was defined as the degree to which attitude manifests itself in the form of temporal persistence, resistance to counter persuasion and predictability of behavior (Petty & Cacioppo, 1986). In more methodological terms, Eaton et al. (2008) extend this definition concluding from a series of empirical studies that AS has both a moderating influence on the effects of attitudes on the behavioral intention, and a direct influence on behavior. A deeper understanding of the definition involves the question whether AS can be regarded as a one-dimensional construct or, if not, which components it includes. Krosnick and Petty (1995) propose a large number of AS indicators, from which Visser et al. (2006) select the following as the most relevant. The importance of an attitude describes the motivation of an individual to activate and use an attitude in certain situations, and to protect it from contrary influences. Earlier studies show that the importance assigned to an attitude is correlated with the importance associated to the object of that attitude (Boninger, Krosnick, Berent, & Fabrigar, 1995). Importance is positively associated with the knowledge the individual has about the object of his or her attitude, which refers to knowledge width, in a mainly quantitative sense, rather than to the depth of that knowledge. Importance is also associated with the accessibility of attitudes, such that stronger attitudes can be more quickly recalled and the individuals do not need long time to reflect about their evaluations; this, in turn, is more likely to happen if the attitudes in question are activated and used more frequently. AS is further associated with the certainty with which the individual can express attitudes, for instance in discussions. A strong attitude will display little ambivalence, meaning that the evaluations that make up the attitude are highly consistent with each other, rather than being positive from some points of view and negative from others. Finally, attitudes display various depths of cognitive elaboration depending on the number and quality of logical links established between the attitude and individual’s related knowledge and experiences. Few studies combine the concept of AS with with technology acceptance models. Kim et al. (2009) add AS to the TAM and address Venkatesh et al.’s (2003) assertion that including attitudes in the TAM does not increase the variance explained by the model (see also Nistor & Heymann, 2010; Teo, 2009). As Kim et al. (2009) show, only strong attitudes mediate the relationship between PE/ EE and the intention to use technology. If attitudes are weak, PE finds itself in a direct relationship with the behavioral intention. From a slightly different point of view, Bhattacherjee and Sanford (2009) propose that two AS components, personal relevance and related expertise moderate the relationship between behavioral intention and use behavior. The later moderator is also present (as “experience” that may be understood as “rich experience,” or “expertise”) in the UTAUT (Venkatesh et al., 2003) and UTAUT2 (Venkatesh et al., 2012). To conclude, there seems to be a consense among researchers that AS has a substantial moderating influence on the paths of current acceptance models. However, the findings are sofar © 2019 British Educational Research Association Attitude strength and educational technology acceptance 5 relatively scarce and little consistent, so that the position of AS within acceptance models—moderator or additional predictor—cannot be stated more precisely. Moreover, the studies of AS and technology acceptance reviewed above are hardly related to educational technologies. Additional research is necessary. Frequent changes of educational technology The wide-scale implementation of educational technology, especially computers and the Internet, is an ever-lasting idea in educational sciences. Hiltz and Turoff wrote as soon as 1978 about “the network nation” and addressed the virtual classroom at the beginning of the 1990s (Hiltz, 1994). In Germany, for instance, the first online seminars were carried out at LudwigMaximilian-Universität München in 1995 (Nistor & Mandl, 1997), on a rudimentary platform developed ad hoc. Ten years later, online seminars with remote students were part of the teaching routine (Nistor, Schnurer, & Mandl, 2005), offered in the framework of the Virtual University of Bavaria (Nistor, 2002). After trying out several self-developed and commercial learning management systems (LMS), Moodle was introduced in 2008, and it is being used in parallel to other LMS (Nistor, 2013). In addition to these changes in the implemented technology, and to regular software updates, additional restrictions and changes have been imposed in May 2018 by the federal data protection act (Bundesdatenschutzgesetz [BDSG], 2018). Today’s users of LMS in higher education are prominently the so called “millennials”, supposed to be familiar with all kind of technologies (Barak, 2018). After more than a decade of Web 2.0 (including, eg, blog platforms like WordPress), we can expect Moodle to be neither the first, nor the most innovative or exciting content or learning management system the millennials use. Given the extensive experience with the Internet and educational technologies, and the frequent technology changes, how strong can actual students’ attitude toward educational technology and LMS be? On the one hand, a change in the educational technology students need to use for their study, eg, the LMS established in their university, may trigger a brief learning process described by the cognitive pattern variability-stability-flexibility (Ionescu, 2017), developing flexible computer skills (Barak & Levenberg, 2016). This in turn may make them open to change (Barak, 2018), and enhance their performance vs. effort expectancies, thus increasing their technology acceptance (Venkatesh et al., 2003, 2012). This conclusion corresponds to the dominant perspective in the technology acceptance research literature (eg, Al-Emran et al., 2018; Khechine et al., 2016; Šumak et al., 2011). On the other hand, being forced to give up an instrument productively used for a while, and to change habits, is a case of forced compliance generating cognitive dissonance. This discomfort can be reduced by a change of attitudes (Festinger & Carlsmith, 1959). Indeed, students may develop a positive attitude toward the new system by self-perception (Bem, 1972), self-determination (Ryan & Deci, 2017), or simply by the reward of being able to work productively (Venkatesh et al., 2012). However, the stronger their attitudes toward technology are, the bigger the discomfort in altering their attitudes will be (Howe & Krosnick, 2017). Hence, frequent technology updates may lead to weaker attitudes, thus to less positive attitudes (Visser et al., 2006), ie, to less technology acceptance. These considerations suggest that AS may be an important acceptance factor in the context of frequently changing technology. However, the role of AS in current technology acceptance models has not been sufficiently clarified, yet. The study presented in the following aimed to provide empirical evidence of this role, thus deepening our understanding of acceptance and suggesting evidence-based implementation strategies of educational technologies. © 2019 British Educational Research Association 6 British Journal of Educational Technology Vol 0 No 0 2019 Research questions Against the background of these theoretical consideration, our study addresses the following research questions. RQ1: What is the relationship between students’ AS and their acceptance of Moodle? As suggested by the existing studies (Howe & Krosnick, 2017; Visser et al., 2006), we expect AS to moderate the relationships between the UTAUT variables, thus extending the findings of Kim et al. (2009) and Bhattacherjee and Sanford (2009). RQ2: Which combinations of AS and acceptance predictors lead to acceptance? The same acceptance predictors may be present in some cases, or absent in others (Agudo-Peregrina et al., 2014; Pynoo & van Braak, 2014). As suggested by Pappas et al. (2019), we seek to go beyond linear models that consider only the presence of acceptance factors, and to find configurations including both the presence and the absence of acceptance factors (Ragin, 2008). Methodology Research design This study is a case study, which is important to state, as we assume that the acceptance factors always depend on technologies and their application contexts, which are different from case to case. Further, we employed a correlational design with one-shot cross-sectional data. Population and sample The study was conducted at Ludwig-Maximilian-Universität München, a large German university with a total of approx. 50 000 students. As mentioned above, the students were accustomed with digital media, including educational technologies, and with repeated changes and updates of technology. At the Faculty of Psychology and Educational Sciences, where the study was conducted, the officially supported and recommended LMS was Moodle. Faculty and students used Moodle in various ways, from repository of learning materials to platform for online seminars with complex instructional designs (Nistor, 2013). Furthermore, Moodle was used in parallel with other LMS, eg, the LSF system for the administration of course participant lists and grades, which also has a repository function used by some instructors to store learning materials. The use of Moodle was recommended but not required, in line with the university regulations that require faculty members to give students traditional, offline alternatives to online learning. Technical support for Moodle was hardly provided, assuming that the usage was intuitive and easy for the students; we were not aware of students’ acute need for technical help. Thus, this case of educational technology use mainly aimed at increasing student performance (high PE in the sense of the UTAUT; Venkatesh et al., 2003, 2012), and it was implemented along with a limited influence from the faculty (moderate SI values). Hence, the case appears different from other cases of LMS use, for instance in business administration, where the most important goal (and thus acceptance factor) is effort minimization (high EE) and, correspondingly, technical help is more intensively provided (eg, Murillo Montes de Oca & Nistor, 2014). The target population consisted of millennials, undergraduate students, from which 251 responded to the invitation to participate and provided complete data. After 15 incomplete data sets and 11 outliers were removed, the final sample consisted of N = 225 participants, 202 female and 23 male, which is a gender distribution typical for the study of Educational Sciences. Measures and instruments Use intention (UI) was regarded as the acceptance indicator, and as a proxy for the technology use behavior, as practiced in the majority of acceptance studies (Khechine et al., 2016; Šumak © 2019 British Educational Research Association Attitude strength and educational technology acceptance 7 et al., 2011). The assessed acceptance predictors were those specified by the UTAUT (Venkatesh et al., 2003, 2012): PE, EE, SI and FC. All acceptance variables were measured using the acceptance questionnaire validated by Venkatesh et al. (2003). To avoid misunderstandings, it must be noted that EE means users’ expectancy that technology reduces their effort; higher EE scores correspond to stronger expectations of reduced effort. To measure AS toward Moodle, an additional subscale was formulated for this study, based on Visser et al.’s (2006) dimensions: knowledge (“I know enough about Moodle to have a clear attitude towards it”), accessibility (“If someone asks me, what I think about Moodle, I can always give a quick answer”) and certainty (“I am quite sure about my attitudes towards Moodle”). All questionare items could be answered using seven point Likert scales (form 1 = low to 7 = high acceptance/AS). Due to poor reliability, FC was disregarded in the further analysis. The other scales prove reliable, with Cronbach’s α between 0.69 and 0.91 (threshold value 0.65), and composite reliability CR between 0.76 and 0.91 (threshold value 0.70); the scales also prove valid, with convergent validity (average variance explained, AVE) between 0.51 and 0.71 (threshold value 0.50), and the AVE square root values all higher than the correlations between variables (discriminant validity). Details are provided in Table 1. In terms of goodness-of-fit, CFI = 0.991 (threshold value 0.95), TLI = 0.975 (threshold value 0.95) and RMSEA = 0.051 (threshold value 0.05) indicate acceptable model fit. Data collection and analysis An invitation to the study was sent by email, along with the informed consent declaration, to all undergraduate students with a major in Educational Sciences. The data were collected anonymously using an online questionnaire. A week after the invitation, a reminder was sent; another week later, the questionnaire was closed and the data downloaded and processed using R (version 3.5.1, with libraries lavaan and psych) and the fs/QCA software (version 2.5). As the fsQCA may be less familiar to the readership than the other analysis methods, here are a few explanations. Basically, the fsQCA terminology makes a difference between outcome, ie, a situation or state of things whose existence or absence is of interest for the research, and conditions, ie, situations of phenomena whose presence or absence may contribute to the occurrence or the absence of the outcome of interest (Ragin, 2008). In the present case, AT, EE, PE and SI obtained during confirmatory factor analysis were treated as conditions of the presence or absence of the outcome UI. The first step in fsQCA is the calibration, ie, the allocation of individual observations or cases in clearly defined sets separated from each other. Here, such a set distinction was made between participants to whom a definite future use intention could be attributed and, reciprocally exclusive, those to whom no future use intention could be attributed. As recommended by Duşa (2018), we used the 0.05, 0.50 and 0.95 percentiles for the complete non-inclusion, cutoff point and complete set-inclusion respectively (Table 2). Next, necessity and sufficiency analyses were conducted, concluded by the minimization procedure. These analyses seek to find the simplest combinations of conditions that lead to the examined outcome. The quality of fsQCA solutions is indicated by coverage and consistency. The former measures the intersection of the outcome with a complete configuration of factors (raw coverage) or with single configurations (unique coverage). Coverage can be regarded as analog to R2 in regression analyses. Consistency describes the agreement in exhibiting the same outcome occurred between cases sharing a factor configuration (Ragin, 2008). © 2019 British Educational Research Association 8 Entire sample PE EE SI FC UI Low AS (N = 126, M = 3.51, SD = 1.01) High AS (N = 125, M = 5.84, SD = 0.65) α CR AVE M SD M SD M SD t df p 0.90 0.91 0.77 0.65 0.69 0.90 0.91 0.76 0.71 0.81 0.71 0.71 0.51 0.45 0.65 4.46 4.96 3.71 1.48 1.46 1.34 4.18 4.51 3.31 1.44 1.45 1.18 4.74 5.42 4.12 1.47 1.33 1.37 −3.02 −5.14 −4.92 248.73 247.45 243.02 0.003 0.000 0.000 4.28 1.65 4.03 1.59 4.54 1.68 −2.48 248.10 0.012 British Journal of Educational Technology © 2019 British Educational Research Association Table 1: Reliability of measures (α = Cronbach’s α; CR = composite reliability; AVE = average variance explained) and descriptive statistics (M = mean value, SD = standard deviation) for the entire sample, and for the AS median split subgroups Vol 0 No 0 2019 Attitude strength and educational technology acceptance 9 Table 2: Percentiles used for data calibration Constructs derived from factor scores Percentile AS EE PE SI UI 0.05 0.50 0.90 −2.51 0.26 2.14 −3.15 0.24 2.03 −1.97 0.13 1.65 −1.94 −0.04 2.75 −3.38 0.46 2.52 Findings RQ1: The relationship between AS and acceptance. The acceptance variables PE, EE, SI and UI and participants’ AS had normally distributed (Kolmogorov-Smirnov test non-significant), moderate values, as displayed in Table 1. For the entire sample, AS significantly predicted PE (β = 0.37, p < .000; R2 = .08), EE (β = 0.42, p < .000; R2 = .18) and SI (β = 0.28, p < .000; R2 = 0.14). PE and SI were significant predictors of the UI (PE: β = 0.31, p < .000; SI: β = 0.28, p < .000), explaining R2 = .29 of the variance in UI. The relationship between EE and UI, was not significant (p = .76) (Figure 1). A median split of the sample by the AS values was performed, which resulted in significantly higher acceptance variable values for the high AS subgroup (Table 1). Subsequently, the acceptance variables and the acceptance model path coefficients were recalculated. For the low AS subgroup (Figure 2a), AS significantly predicted PE (β = 0.26, p < .002; R2 = .08), EE (β = 0.30 p < .000; R2 = .10), and SI (β = 0.33, p < .000; R2 = .11). Only PE and SI were significant predictors of the UI (PE: β = 0.26, p < .001; SI: β = 0.37, p < .000), explaining R2 = .26 of the variance in UI. The relationship between EE and UI was, again, not significant (β = −0.11, p = .15). For the high AS subgroup (Figure 2b), AS predicted only EE (β = 0.34, p < .000, R2 = .12), the relationships AS-PE (β = 0.16, p = .07) and AS-SI (β = 0.14, p = .13) were not significant. PE, EE and SI were significant predictors of the UI (PE: β = 0.26, p < .000; EE: β = 0.16, p < .001; SI: β = 0.25, p < .000), explaining R2 = .32 of the variance in UI. RQ2: Acceptance-relevant combinations of AS and acceptance. As suggested by Pappas et al. (2019), we sought to go beyond linear models that consider only the presence of acceptance factors, and to find configurations including both the presence and the absence of acceptance factors, or combinations thereof (Ragin, 2008). The Table 3 presents the minimized solutions for two possible outcomes, the definite intention to use, and, respectively, the complete absence of intention to use the system. The best coverage (90%) of cases of definite use intention was given by the addition of the participant sets who either have a high PE, or perceive strong SI, or experience a combination of low EE conjoined with low AS. The overall solution consistency revealed that 72% of the solution was found when the participants expressed a definite positive use intention. 5ð   $WWLWXGH 6WUHQJWK 3HUIRUPDQFH ([SHFWDQF\  5ð  5ð   (IIRUW ([SHFWDQF\  6RFLDO ,QIOXHQFH  8VH ,QWHQWLRQ 5ð   Figure 1: Model path coefficients and explained variance for the entire sample (***p < .000) © 2019 British Educational Research Association 10 British Journal of Educational Technology Vol 0 No 0 2019 5ð  (a)  $WWLWXGH 6WUHQJWK 3HUIRUPDQFH ([SHFWDQF\  5ð  5ð   (IIRUW ([SHFWDQF\  8VH ,QWHQWLRQ 5ð   6RFLDO ,QIOXHQFH  (b) Figure 2: Model path coefficients and explained variance for the low (a) versus high (b) AS subsamples (***p < .000, **p < .01, †p < .10) Table 3: Configurations of acceptance factors leading to positive and, respectively, negative outcome (minimized solutions, ● = present factor; Ø = absent factor) UI present Type of outcome (UI) Solutions AS (attitude strength) PE (performance expectancy) EE (effort expectancy) SI (social influence) Raw coverage Unique coverage Consistency Overall solution coverage Overall solution consistency 1a 2a 3a 1b 2b 3b Ø ● 0.79 0.11 0.79 0.90 0.72 UI absent ● 0.80 0.07 0.77 Ø ● Ø 0.46 0.01 0.81 0.75 0.03 0.66 0.86 0.74 0.79 0.08 0.68 Ø 0.60 0.03 0.68 Conversely, for the definite absence of use intention, the best coverage (86%) was given by the addition of the participant sets with either low-effort expectancy, or low-performance expectancy or lack of perceived social influence. The overall solution consistency for the lack of use intention was 74%. Discussion Confirming our expectations, we found moderate and normally distributed values of AS and the acceptance variables PE and EE, and lower values of SI (Kim et al., 2009). Furtheron, we could verify the basic UTAUT model (Venkatesh et al., 2003, 2012) consisting of PE, EE, SI and UI. However, EE was not a significant predictor of UI, which we interpret as a particularity of the examined case, rather than evidence questioning the UTAUT validity. Altogether, the basic UTAUT explained nearly one third of the variance in UI, which is in line with previous research (Al-Emran et al., 2018; Khechine et al., 2016; Šumak et al., 2011). As suggested by social © 2019 British Educational Research Association Attitude strength and educational technology acceptance 11 psychological research (Bhattacherjee & Sanford, 2009; Eaton et al., 2008; Howe & Krosnick, 2017; Visser et al., 2006), we found significant direct relationships between AS and all acceptance predictors PE, EE and SI. Two paths indirectly connected AS with UI: firstly, the path ASPE-UI that we interpret as an expression of students’ performance orientation; secondly, the path AS-SI-UI that we interpret as students’ compliance with the instruction and conformity with the existing learning culture (eg, Scherer et al., 2015). The non-significant AS-EE-UI path calls for a more careful interpretation, if any. The median split reveals that the picture remains the same if AS is low, ie, for weak attitudes. In other words, it is weak attitude toward Moodle that may be manifested twofold—firstly, as PE related to the academic study (which can be briefly put in a student statement like “I am fine with any technology that helps me achieve my study goals”); secondly as SI in the sense of compliance with instruction (“I am fine with any technology recommended by the instructor”). This may be the case when Moodle usage has become routine to the students, who then concentrate their attention and activity on the learning contents and processes supported by the platform, rather than on the platform itself. High AS reverses the picture, making the AS-EE-UI path significant, and the other two somewhat past the significance level, ie, non-significant. This suggests that strong attitudes toward Moodle manifest themselves as a strong expectation that Moodle reduces their effort to study. This might happen when students encounter technical problems that break the routine and initiate potentially effortful trouble shooting—that might as well annoy the students and trigger strong negative attitudes (Orlando, 2014). This may also be a credible explanation why, from all three considered acceptance predictors it is EE from which technology-related AS explains the largest amount of variance (up to 18%). Altogether, the UTAUT path analysis of the examined case could be briefly put in a student statement like “I am fine with any technology, as long as it doesn’t make trouble, so that I can concentrate on my study”—which we find perfectly credible. The fsQCA model confirms the importance of the acceptance predictors PE and SI for this case: no matter what the other factors are, any of them will impose their value (present or absent) to the overall acceptance (present or absent). This is consistent with numerous previous findings based on regression and SEM (Al-Emran et al., 2018; Khechine et al., 2016; Šumak et al., 2011). As for EE, the combination of present EE and absent AS will lead to Moodle use intention, as well. This case may be infrequent (as indicated by the low unique coverage of the particular solution), but it is a clearly occurring configuration of acceptance (as indicated by the high consistency)—that can be, as well, expressed in the statement “I am fine with any technology, as long as it doesn’t make trouble, so that I can concentrate on my study.” Comparing the SEM and fsQCA results, we observe that, while the SEM may only suggest the combination of high EE and high AS as possibly conducive to acceptance, this suggestion can only be elicited by a statistical artifact (median split) that requires cautious interpretation of results (and nonetheless researcher’s imagination). Moreover, the SEM hardly provides any hint as to how frequent the AS-EE configuration occurs alone (unique coverage) or in combination with others (raw coverage), which fsQCA does. Thus, fsQCA, while consistent with the SEM, offers a sharper lens and a clearer picture of the phenomenon. The still unanswered question whether AS should be regarded as a direct predictor of acceptance variables or as an UTAUT moderator (Bhattacherjee & Sanford, 2009; Howe & Krosnick, 2017) becomes futile when fsQCA is applied. Consequences The validity and generalizability of the presented research is not free from limitations. As noted at several places above, this was a single case of technology use for specific purposes; to overcome this limitation, more cases of technology use with different contexts and goals should be © 2019 British Educational Research Association 12 British Journal of Educational Technology Vol 0 No 0 2019 examined in future research. Furthermore, this field research assumed that the number of technology changes prior to the study was a constant for all participants; there is a need for laboratory research that includes the manipulation of technology changes, and verifies its effects on AS. Finally, this study was based on a simple quantitative operationalization of technology acceptance, as already criticized by Benbasat and Barki (2007); the study should be complemented by qualitative research aimed at a deeper insight in users’ goals of technology use and the corresponding evaluations that make up their attitudes. Nevertheless, based on the theoretical considerations and empirical findings presented above, we suggest at least three implications. Firstly, low AS may occur in response to frequent technology change, which may be regarded as a means to reduce the cognitive dissonance resulting from forced compliance (Festinger & Carlsmith, 1959) with technological constraints that affect habits (Venkatesh et al., 2012). Secondly, AS will probably be positively related to all acceptance predictors and UI, and moderate the relationships between them. Thus, low AS may affect technology acceptance. Thirdly, before looking at learners’ intention to use a technology, it should be first clarified whether the learners have a clearly defined attitude toward the technology at all, and subsequently how strong that attitude is. In this sense, current technology acceptance models may be reconsidered. While the current models are not necessarily invalid, their validity should be extended by adding AS. Furthermore, additional research is needed to elucidate the relationship between frequent technology changes and AS, and to extend these findings to other educational technology settings. Research on technology-related AS should be conducted involving larger samples of students, and objective measures of technology use behavior such as log data. As an important methodological innovation, fsQCA can provide a clearer insight in the acceptance phenomenon by indicating configurations of factors that lead to positive or negative outcome (in this case, acceptance or, more specifically, UI), including the presence and absence of those factors. Thus, the fsQCA results extend the results of regression or SEM. For the implementation of educational technology in higher education, technology acceptance models extended by AS in combination with acceptance configurations identified by fsQCA may point at various situations within the setting in which technology users may accept and use a technology to a higher or lower degree. Thus, acceptance research may move beyond the focus on individual cognitions or characteristics of users indiscriminately grouped together as one ideal user. Elaborating on different patterns, or situations, of use can lead to targeted educational interventions that enhance acceptance behaviors and, in the case of educational technology, that enhance learning. Statements on open data, ethics and conflict of interest Readers who wish to examine the data can contact the first author. All procedures performed in this study follow the ethical standards of the of the institutional research committee. Authors have no conflict interest. References Agudo-Peregrina, Á. F., Hernández-García, Á., & Pascual-Miguel, F. J. (2014). Behavioral intention, use behavior and the acceptance of electronic learning systems: Differences between higher education and lifelong learning. Computers in Human Behavior, 34, 301–314. https://doi.org/10.1016/j.chb.2013.10.035 Ajzen, I., & Fishbein, M. (2000). Attitudes and the attitude-behavior relation: Reasoned and automatic processes. European Review of Social Psychology, 11(1), 1–33. https://doi.org/10.1080/14792779943000116 © 2019 British Educational Research Association Attitude strength and educational technology acceptance 13 Al-Emran, M., Mezhuyev, V., & Kamaludin, A. (2018). Technology acceptance model in M-learning context: A systematic review. Computers & Education, 125, 389–412. https://doi.org/10.1016/j.compe du.2018.06.008 Barak, M. (2018). Are digital natives open to change? Examining flexible thinking and resistance to change. Computers & Education, 121, 115–123. https://doi.org/10.1016/j.compedu.2018.01.016 Barak, M., & Levenberg, A. (2016). A model of flexible thinking in contemporary education. Thinking Skills and Creativity, 22, 74–85. https://doi.org/10.1016/j.tsc.2016.09.003 Bem, D. J. (1972). Self-perception theory. Advances in Experimental Social Psychology, 6, 1–62. https://doi. org/10.1016/S0065-2601(08)60024-6 Benbasat, I., & Barki, H. (2007). Quo vadis, TAM? Journal of the Association for Information Systems, 4(3), 211–218. Bhattacherjee, A., & Sanford, C. (2009). The intention–behaviour gap in technology usage: The moderating role of attitude strength. Behaviour & Information Technology, 28(4), 389–401. https://doi. org/10.1080/01449290802121230 Boninger, D. S., Krosnick, J. A., Berent, M. K., & Fabrigar, L. R. (1995). The causes and consequences of attitude importance. In J. A. Krosnick & R. E. Petty (Eds.), Attitude strength: Antecedents and consequences (pp. 159–189). New York, NY: Psychology Press. Bundesdatenschutzgesetz [German federal data protection act]. (2018). Retrieved from https://www.geset ze-im-internet.de/bdsg_2018/index.html Davis, F. D. (1985). A technology acceptance model for empirically testing new end-user information systems: Theory and results (Unpublished doctoral dissertation). Boston, MA: Massachusetts: Institute of Technology. Retrieved from https://dspace.mit.edu/bitstream/handle/1721.1/15192/14927137-MIT. pdf Duşa, A. (2018). QCA with R: A comprehensive resource. New York, NY: Springer. Eaton, A. A., Majka, E. A., & Visser, P. S. (2008). Emerging perspectives on the structure and function of attitude strength. European Review of Social Psychology, 19(1), 165–201. https://doi.org/10.1080/10463 280802383791 Festinger, L., & Carlsmith, J. M. (1959). Cognitive consequences of forced compliance. The Journal of Abnormal and Social Psychology, 58(2), 203–210. https://doi.org/10.1037/h0041593 Hiltz, R. (1994). The virtual classroom: Learning without limits via computer networks. Norwood, NJ: Ablex Publishing. Howe, L. C., & Krosnick, J. A. (2017). Attitude strength. Annual Review of Psychology, 68, 327–351. https:// doi.org/10.1146/annurev-psych-122414-033600 Hur, H. J., Lee, H. K., & Choo, H. J. (2017). Understanding usage intention in innovative mobile app service: Comparison between millennial and mature consumers. Computers in Human Behavior, 73, 353–361. https://doi.org/10.1016/j.chb.2017.03.051 Ionescu, T. (2017). The variability-stability-flexibility pattern: A possible key to understanding the flexibility of the human mind. Review of General Psychology, 21(2), 123–131. https://doi.org/10.1037/gpr00 00110 Khechine, H., Lakhal, S., & Ndjambou, P. (2016). A meta-analysis of the UTAUT model: Eleven years later. Canadian Journal of Administrative Sciences/Revue Canadienne des Sciences de l’Administration, 33(2), 138– 152. https://doi.org/10.1002/cjas.1381 Kim, Y. J., Chun, J. U., & Song, J. (2009). Investigating the role of attitude in technology acceptance from an attitude strength perspective. International Journal of Information Management, 29(1), 67–77. https://doi. org/10.1016/j.ijinfomgt.2008.01.011 Krosnick, J. A., & Petty, R. E. (Eds.). (1995). Attitude strength: Antecedents and consequences. New York, NY: Psychology Press. Martí-Parreño, J., Galbis-Córdova, A., & Miquel-Romero, M. J. (2018). Students’ attitude towards the use of educational video games to develop competencies. Computers in Human Behavior, 81, 366–377. https:// doi.org/10.1016/j.chb.2017.12.017 © 2019 British Educational Research Association 14 British Journal of Educational Technology Vol 0 No 0 2019 Murillo Montes de Oca, A., & Nistor, N. (2014). Non-significant intention-behavior effects in educational technology acceptance: A case of competing cognitive scripts? Computers in Human Behavior, 34, 333– 338. https://doi.org/10.1016/j.chb.2014.01.026 Nistor, N. (2013). Etablierte Lernmanagementsysteme an der Hochschule: Welche Motivation ist dabei wünschenswert? In C. Bremer & D. Krömker (Eds.), E-Learning zwischen Vision und Alltag. Zum Stand der Dinge (pp. 181–190). Münster, Germany: Waxmann. Nistor, N. (2014). Exploring non-significant intention effects on educational technology use behavior. Editorial of the special issue “Educational technology acceptance: Explaining non-significant intention-behavior effects”. Computers in Human Behavior, 34, 299–300. https://doi.org/10.1016/j. chb.2014.02.052 Nistor, N. (2002). Die Virtuelle Hochschule Bayern: Aktueller Stand und Zukunftsperspektiven. In G. Bachmann, O. Haefeli, & M. Kindt (Hrsg.): Campus 2002: Die virtuelle Hochschule in der Konsolidierungsphase (S. 168–175). Münster, Germany: Waxmann. Nistor, N., & Heymann, J. O. (2010). Reconsidering the role of attitude in the TAM. An answer to Teo (2009a). British Journal of Educational Technology (Colloquium), 41(6), E142–E145. https://doi. org/10.1111/j.1467-8535.2010.01109.x Nistor, N., & Mandl, H. (1997). Lernen in Computernetzwerken: Erfahrungen mit einem virtuellen Seminar. Unterrichtswissenschaft, 25(1), 19–33. Nistor, N., Schnurer, K., & Mandl, H. (2005). Akzeptanz, Lernprozess und Lernerfolg in virtuellen Seminaren: Wirkungsanalyse eines problemorientierten Seminarkonzepts. MedienPädagogik, 5(2). Retrieved from http://www.medienpaed.com/article/view/212/212 Orlando, J. (2014). Veteran teachers and technology: change fatigue and knowledge insecurity influence practice. Teachers and Teaching Theory and Practice, 20(4), 427–439. https://doi.org/10.1080/13540 602.2014.881644 Pappas, I. O., Giannakos, M. N., & Sampson, D. G. (2019). Fuzzy set analysis as a means to understand users of 21st-century learning systems: The case of mobile learning and reflections on learning analytics research. Computers in Human Behavior, 92, 646–659. https://doi.org/10.1016/j.chb.2017.10.010 Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In R. E. Petty & J. T. Cacioppo (Eds.), Communication and persuasion (pp. 1–24). New York, NY: Springer. https://doi. org/10.1007/978-1-4612-4964-1_1 Pynoo, B., & van Braak, J. (2014). Predicting teachers’ generative and receptive use of an educational portal by intention, attitude and self-reported use. Computers in Human Behavior, 34, 315–322. https://doi. org/10.1016/j.chb.2013.12.024 Ragin, C. C. (2008). Redesigning social inquiry: Fuzzy sets and beyond. Chicago, IL: University of Chicago Press. Ryan, R. M., & Deci, E. L. (2017). Self-determination theory: Basic psychological needs in motivation, development, and wellness. New York, NY: Guilford. Sanford, C., & Oh, H. (2010). The role of user resistance in the adoption of a mobile data service. Cyberpsychology, Behavior, and Social Networking, 13(6), 663–672. https://doi.org/10.1089/ cyber.2009.0377 Scherer, R., Siddiq, F., & Teo, T. (2015). Becoming more specific: Measuring and modeling teachers’ perceived usefulness of ICT in the context of teaching and learning. Computers & Education, 88, 202–214. https://doi.org/10.1016/j.compedu.2015.05.005 Schoonenboom, J. (2012). The use of technology as one of the possible means of performing instructor tasks: Putting technology acceptance in context. Computers & Education, 59(4), 1309–1316. https://doi. org/10.1016/j.compedu.2012.06.009 Šumak, B., Heričko, M., & Pušnik, M. (2011). A meta-analysis of e-learning technology acceptance: The role of user types and e-learning technology types. Computers in Human Behavior, 27(6), 2067–2077. https://doi.org/10.1016/j.chb.2011.08.005 Teo, T. (2009). Is there an attitude problem? Reconsidering the role of attitude in the TAM. British Journal of Educational Technology, 40(6), 1139–1141. © 2019 British Educational Research Association Attitude strength and educational technology acceptance 15 Venkatesh, V., & Bala, H. (2008). Technology acceptance Model 3 and a research agenda on interventions. Decision Sciences, 39(2), 273–315. Venkatesh, V., & Davis, F. (2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 45, 186–204. Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS Quarterly, 27(3), 425–478. Venkatesh, V., Thong, J. Y., & Xu, X. (2012). Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Quarterly, 36(1), 157–178. Visser, P. S., Bizer, G. Y., & Krosnick, J. A. (2006). Exploring the latent structure of strength-related attitude attributes. Advances in Experimental Social Psychology, 38, 1–67. Wang, P. (2010). Chasing the hottest IT: Effects of information technology fashion on organizations. MIS Quarterly, 34(1), 63–85. © 2019 British Educational Research Association