The In ﬂ uence of Values on E-learning Adoption

As technology continues to pervade our lives, the in ﬂ uence of culture on technology adoption is of signi ﬁ cant interest to researchers. However, culture, as a group-level construct may not give meaningful results when related to individual-level adoption. Although culture has been integrated into technology adoption models, values are the individual-level representation of culture, and are more appropriate to include in technology adoption models. There have been few studies attempting to explore the in ﬂ uence of values on adoption models, and none within the sphere of digital education. The purpose of this exploratory study is to integrate values with technology adoption models and apply the novel conceptual model to the context of digital education. In this study we investigate the in ﬂ uence of individual-level values on the adoption of e-learning by workers in The Gambia and the UK. Using the Uni ﬁ ed Theory of the Acceptance and Use of Technology (UTAUT2) as a base model, we integrate values relating to conservation of the status quo and self-enhancement from Schwartz's Theory of Human Values. Taking this approach, we develop and introduce the Values-Enhanced Technology Adoption (VETA) model. We tested the VETA model on the adoption of e-learning by workers in The Gambia and the UK. Empirical results demonstrated the in ﬂ uence of self-enhancement values in the model via social in ﬂ uence, price value and performance expectancy. The UTAUT2 base model was partially validated in that performance expectancy, price value and habit primarily in ﬂ uenced worker intention to use e-learning. We conclude that VETA will be a useful model to researchers studying technology adoption


Introduction
In recent years, there have been significant investments in digital learning technologies across all sectors. Despite such investments, learning technology might be poorly adopted, discontinued or rejected by learners (Adelé & Brangier, 2013;Recker, 2016), therefore learner motivation and engagement remain topics of interest for researchers (Jung & Lee, 2018;Zhu, Sari, & Lee, 2018). Individual acceptance of technology has been investigated within the information systems literature, indicating that intention to use technology is predicted by user perceptions of technology attributes, specifically usefulness and ease of use (Davis, Bagozzi, & Warshaw, 1989).
Technology adoption models generally explain user intention as a function of user perceptions of attributes of the target technology within the context of use (Venkatesh & Bala, 2008;Venkatesh & Davis, 2000). The latest iterations of technology adoption models include individual consumer preferences (price value, habit and intrinsic motivation) (Venkatesh, Thong, & Xu, 2012) and context (social factors and physical infrastructure) (Taylor & Todd, 1995;Venkatesh et al., 2012;   . In the context of technology adoption models, incorporating social factors requires some consideration of culture (an aggregate, group-level framework that represents societal views), or values (an individual-level framework of internally held, enduring beliefs) (Hofstede, Hofstede, & Minkov, 2010;Rokeach, 1973;Schwartz, 1992). Few studies have integrated culture or values into the technology acceptance framework (Srite & Karahanna, 2006), using group-level cultural dimensions: the effects of individual-level values on individual-level decision-making remains an area for further investigation. This paper proposes that individual-level values influence user perception of the relative importance of technology, context or consumer-focussed attributes that lead to the intention to use technology. This study surveys workers across two contexts (The Gambia and the UK) to gather technology adoption and values data to determine the influence of these values on the adoption model derived from UTAUT2. The research objective of this study is to extend the UTAUT2 model to incorporate values at the individual level. This study contributes to the literature by being amongst the first to incorporate Schwartz's values framework into technology adoption literature, to investigate the influence of values on elearning adoption, and to develop and validate the Values-Enhanced Technology Adoption (VETA) model.

Literature review
The adoption of e-learning depends on a learner's decision to use the technology, expressed in the literature as behavioural intention . The Technology Acceptance Model (TAM), explains behavioural intention as a product of user attitude toward technology attributes: perceived usefulness and perceived ease of use . Venkatesh et al. empirically compared and consolidated the acceptance literature producing the Unified Theory of the Acceptance and Use of Technology (UTAUT and UTAUT2) which increased the explanatory power of prior technology acceptance models (Venkatesh et al., 2003(Venkatesh et al., , 2012. Although TAM has been used to explain learner adoption of technology-enhanced learning (Abbasi, Chandio, Soomro, & Shah, 2011;Ong, Lai, & Wang, 2004;Roca & Gagné, 2008), extended models such as UTAUT (Khechine, Pascot, & Bytha, 2014;Sumak, Polancic, & Hericko, 2010) and UTAUT2 (El-Masri & Tarhini, 2017;Raman & Don, 2013;Yang, 2013) have demonstrated the role of social and physical context, and consumer judgement in determining learner behaviour with educational technology.
User attitude, based on prior experience or an individual's worldview or culture, is linked with values, which may influence perceptions of technology or contextual attributes that help form decisions (Fishbein & Ajzen, 1975b;Rokeach, 1975). Culture is a multi-layered, multi-dimensional construct that represents an aggregate value system shared by individuals within a group (Hofstede et al., 2010;Rokeach, 1968;Schwartz, 2012). (Taras, Rowney, & Steel, 2009). Values are internally held beliefs about preferred endstates or outcomes that transcend situation and may act as a standard of behaviour (Rokeach, 1973). At an individual level, values explain the motivational foundation of attitude and behaviour, while the prioritisation of value structures at a group level help determine culture (Boer & Fischer, 2013;Schwartz, 1994a). Attitudes are formed towards an object applied in a scenario, and differ as values priorities shift even if the object and scenario are kept constant (Fishbein & Ajzen, 1975a). The premise that an attitude leads to a discriminatory response links values and technology acceptance theories.
The literature contains some studies that have investigated the influence of espoused culture (as a proxy for individual-level values) on user behaviour with technology (Srite, 2006;Srite & Karahanna, 2006;Tams, Thatcher, & Srite, 2012), with comparative studies in European (Göğüş, Nistor, & Lerche, 2012;Nistor et al., 2014;Nistor, Göğüş, & Lerche, 2013) and Middle-Eastern (Tarhini, Hone, & Liu, 2013;Tarhini, Hone, Liu, & Tarhini, 2017) e-learning applications. These studies generally agree that the social environment provides a mechanism to link subjective norm to intention, with some agreement that collectivist cultures prefer to maintain normative stability. These studies, however, do not include African cultures, nor do they use theoretically derived values at an individual level to explore learner perceptions, relying instead on espoused national culture arbitrarily demarcated at geographical borders (McSweeney, 2002;Schaffer & Riordan, 2003).
Prior investigation on culture and values in technology acceptance models has used the empirically derived group-level dimensions of culture proposed by Hofstede, operationalised by Dorfman and Howell (Dinev, Goo, Hu, & Nam, 2009;El-Masri & Tarhini, 2017;Hoehle, Zhang, & Venkatesh, 2015;McCoy, Galletta, & King, 2005;Srite & Karahanna, 2006). These studies collectively have found that Hofstede's group-level cultural constructs, such as power distance and individualism, espoused at an individual level, can moderate the influence of adoption factors, such as subjective norms and perceived usefulness, on behavioural intention to use technology. However, these studies do not always agree on which dimensions produce an effect and demonstrate that effects differ across different countries and population groups (Nistor et al., 2014;Udo, Bagchi, & Kirs, 2012). The impact of Hofstede's cultural dimensions on the landscape of cross-cultural literature is significant, but not without critique (McSweeney, 2002). However, using group-level constructs such as culture to explain individual-level decisions potentially risks the ecological fallacy: an individual-level proxy for culture must be sought to integrate with decision models that operate at the individual level (Brewer & Venaik, 2014).
Schwartz's theoretically derived values model (Table 1) represents an individual's world-view that lies at the heart of culture (Rokeach, 1968(Rokeach, , 1973Schwartz, 1994b;Schwartz et al., 2012). Schwartz's theory of human values operationalizes values as a continuum (Table 1), applicable at an individual-level, as antecedents to attitude that are non-specific in how they relate to a target behaviour (Rokeach, 1973). Relatively few studies have attempted to integrate Schwartz's values with decision models (Ahmad & Sun, 2018;Diddi & Niehm, 2017;Goncalves, Oliveira, & Cruz-Jesus, 2018;Grigoryan, Lebedeva, & Breugelmans, 2018;Pahnila, Siponen, Myyry, & Zheng, 2011;Seddig & Davidov, 2018). These studies find that individual-level values and values dimensions can influence a range of attitudes, from perceptions of risk to attitudes towards interpersonal violence. These studies also demonstrate a difference between the approach taken by researchers who implement Hofstede's dimensions as moderators of attitude-behaviour relationships (Srite & Karahanna, 2006;Tarhini et al., 2017), and the approach taken by researchers who implement Schwartz's values as predictors of attitudes or intention (Diddi & Niehm, 2017;Grigoryan et al., 2018). This study extends the UTAUT2 for elearning use intention, to form the Values-Enhanced Technology Adoption (VETA) model (Fig. 1).

Performance expectancy (PE)
From the primary literature on technology acceptance , meta-analyses (Dwivedi, Rana, Chen, & Williams, 2011), and literature specific to e-learning (El-Masri & Tarhini, 2017;Šumak, Heričko, & Pušnik, 2011), perceptions of usefulness and ease of use are important in predicting behavioural intention to use e-learning. In UTAUT, performance expectancy is the 'degree to which an individual believes that system use will yield gain in work performance' and is typically the strongest predictor of intention. For workplace e-learning to be perceived as useful, learners judge their ability to achieve outcomes in their working environment (Chau, 2010). Therefore, consistent with TAM, UTAUT and UTAUT2, we include performance expectancy (PE) as a predictor of behavioural intention (BI).

Effort expectancy (EE)
TAM and UTAUT have been used to explore the acceptance of e-learning by students and teachers in higher education settings, as well as the acceptance of vocational e-learning by employees in business settings. However, research has been conducted Table 1 Schwartz's values framework .

Value Dimension Value Conceptual Definition & Sub-Types
Openness to Change Following one's own intellectual interests Self-Direction Self-direction of action and of thought. Priority of independent thought and action. Stimulation Priority of variety and excitement in life.

Hedonism
Priority of pleasure and gratification.

Self-Enhancement
Priority of personal interests over others Achievement Priority of personal success through the demonstration of competence. Power Power over resources or dominance over people. Priority of social status, prestige and control over people and resources.

Conservation
Preserving the status quo Security Personal and societal security. Priority of safety, harmony and stable social relationships. Tradition Priority of existing customs and ideals.

Conformity
Interpersonal and rule-based conformity. Priority of restraint in actions and impulses.

Self-Transcendence
Commitment to the welfare of others Benevolence Priority of the welfare of others. Universalism Priority of the welfare of the natural environment and wider society. predominantly in developed countries with mixed results: the relevance of perceived ease of use is greatest for vocational settings; the applicability of the model in developing countries is still relatively under-tested (Mbarika, Meso, Musa, Kah, & Amougou, 2007;Raman et al., 2014;Thomas, Singh, & Gaffar, 2013;Šumak et al., 2011). Effort expectancy is analogous to ease of use (TAM) and represents the 'degree to which system use is free from effort'. From the TAM , effort expectancy is expected to predict performance expectancy as users perceive a technology to be more useful when it is easy to use. Therefore, we include effort expectancy as both a predictor of performance expectancy (PE) and behavioural intention (BI).

Social influence (SI)
Social influence represents the pressure of subjective norm defined as the 'perception of group influence on an individual's decision'. In an organisation, users may perceive a technology to be more useful in helping to achieve job-related goals if referent others in the organisation, such as managers, endorse the use of such technology. The predictive influence of social factors on performance expectancy has been made in prior literature (Abdullah & Ward, 2016;Dwivedi, Rana, Jeyaraj, Clement, & Williams, 2017;Venkatesh et al., 2003). Similar to performance expectancy, research in Africa has found varying effects from social influence on behavioural intention (Kolog, 2015;Mtebe & Raisamo, 2014). Across the contexts of this study, learners who perceive e-learning as endorsed by referent others within their social environment are more likely to intend to use e-learning. Therefore, consistent with UTAUT and UTAUT2, the VETA model includes social influence (SI) as a predictor of behavioural intention (BI).

Hedonic motivation (HM)
In the context of e-learning, intrinsic, or hedonic, motivation can be linked to learner enjoyment, playfulness with e-learning, learning strategy, the flow of the learning experience, and engagement, as well as the success of online learning (Barak, Watted, & Haick, 2016 (Saadé, Nebebe, & Mak, 2009). Acknowledging Ryan and Deci's Self-Determination Theory, there is likely to be a continuum of motivation for learners to use e-learning technology from purely extrinsic to purely intrinsic, with the reality likely to be a complex mixture of both (Ryan & Deci, 2000). However, e-learning courses that are engaging, well designed and interesting are likely to be perceived as useful by e-learners (Roca & Gagné, 2008). Across the contexts of this study, learners who perceive e-learning as intrinsically motivating are likely to perceive e-learning as useful; and, learners who are intrinsically motivated are more likely to intend to use e-learning, therefore, we include hedonic motivation (HM) in the VETA model, as a predictor of performance expectancy (PE) and behavioural intention (BI).

Price value (PV)
Price value is linked to a user making a cost-benefit decision, therefore a user may deem technologies more useful if the benefits are greater relative to the cost incurred (Venkatesh et al., 2012;Zeithaml, 1988). Although price value is often excluded from, or is found non-significant in studies where applications exclude direct monetary cost to users (El-Masri & Tarhini, 2017;Lewis, Fretwell, Ryan, & Parham, 2013;Raman & Don, 2013), for workplace learning, non-monetary costs may include time and effort required to access e-learning around work priorities, and the benefits can be linked to the extrinsic outcomes from learning, linking overall learning value to performance expectancy (Ain, Kaur, & Waheed, 2016). Therefore, we include price value in the VETA model applied to e-learning in this study, as a predictor of performance expectancy (PE) and behavioural intention (BI).

Habit (HAB)
Established routines in the workplace, for finding information and learning job-related skills can be important in determining future actions with respect to achieving work-related goals and participating in work-related initiatives, such as e-learning (Verplanken & Aarts, 1999;Wood, 2017). Prior behaviours, perceived to be automatic, can result in initial adoption and prolonged use (Kim & Malhotra, 2005;Limayem et al., 2007;Venkatesh et al., 2012), or can provide a barrier to new initiatives (Labrecque et al., 2017). Therefore, consistent with UTAUT2, we include habit (HAB) in the VETA model, as a predictor of behavioural intention (BI).

The Values-Enhanced Technology Adoption (VETA) model
A learner's perception of the technology or contextual attributes of a workplace e-learning course can be linked with the priority A. Mehta, et al. Computers & Education 141 (2019) 103617 that the learner places on their own interest in self-enhancement, and the priority they place on preserving the organisational framework around them (conservation). Learning, irrespective of whether technology is involved, is likely to be linked with an individual's priority for self-enhancement. E-learning is intended to enhance individual-level performance, and therefore intention to use e-learning should be predicted by a learner's preference for self-enhancement (Veiga, Floyd, & Dechant, 2001). Learner priority on achievement through course completion, and on the extrinsic outcomes that arise from achievement, such as increased status, perceived knowledge and career opportunity are linked to the values of achievement and power within the self-enhancement dimension. With respect to workplace e-learning, values related to Schwartz's conservation dimension are likely to be important in determining learner opinion. Values in the openness-to-change dimension, such as hedonism, are likely to link to learner perception of enjoyment in learning. Values in the self-transcendence domain link to perceptions of society and the natural environment, which are unlikely to be relevant to individual-level adoption of technology such as e-learning, and therefore not included in the VETA model for e-learning adoption, although may be considered when applying the model to technologies that have societal impact. Definitions of Schwartz's values are shown in Table 1. This study theorises that an individual's values will predict their perception of technological, contextual or personal attributes that lead them to make decisions about using technology and are therefore best placed as antecedents to those factors in a structural model. The Values-Enhanced Technology Adoption (VETA) model is shown in Fig. 1.

Conservation values (security (SE), tradition (TR), and conformity (CO))
Individuals expressing high conservation values, both interpersonal and compliance, are likely to place great importance on meeting obligations within their social and organisational spheres including meeting the expectations of significant influencers in their social environment, both in terms of peer and organisational influence (Schwartz, 1992). Conformity in a workplace scenario might arise through following rules laid down by the organisation in terms of mandatory or endorsed training programmes. The value of security contains subtypes of personal security (personal health, safety, wellbeing) and societal security (stable social order) : for vocational training, it can be expected that workers complete basic training delivered via technology to safeguard their future employment against threat of competition from peers and through maintaining competence, therefore individuals who place importance on personal security will perceive educational technology to be useful to their job performance. The value of tradition indicates an individual's priority on existing paradigms within their organisation (Schwartz, 1992). Workers prioritising the importance of existing organisational norms that relate to learning, development and career progression are likely to perceive the completion of workplace e-learning as worthwhile and endorsed by their normative framework. Workers will perceive vocational e-learning as useful to their career objectives, endorsed by referent others in their organisational hierarchy and worth their time and effort to complete. Therefore, we include conservation-type values (security, tradition and conformity) as predictors of performance expectancy and social influence.
3.3.2. Self-enhancement and openness-to-change values (achievement (AC), power (PO) and hedonism (HE)) Workplace learning, with or without technology assistance, can be expected to link with self-enhancement values. The concept of achievement includes the demonstration of competence by the standards of one's reference groups and the pursuit of personal success (Schwartz, 1994a). The inherent worth of e-learning is linked to self-enhancement: learners use content to generate knowledge which improves their understanding, their social status, achievement within standardised testing, and potentially influences their extrinsically-rewarded job performance, such as career or promotion benefits Venkatesh & Davis, 2000). Considering costs and benefits within the concept of price value, benefits of learning technologies may be linked to self-enhancement outcomes, while costs may be linked to context (Ain et al., 2016). The concept of dominance over resources and people are linked to both formal and informal status in an organisation: progression within the formal organisational hierarchy typically brings increased resource responsibility, as well as managerial responsibility over others; increased social status also brings informal power through perceived expertise . However, considering achievement-related outcomes in the light of perceived learner value (Ain et al., 2016), it is expected that learners will consider learning outcomes relative to the effort incurred to access learning materials. The value of hedonism overlaps with both the dimensions of self-enhancement and openness-to-change and can be considered as part of either dimension on the continuum of values proposed by Schwartz. Hedonism, incorporates intrinsic motivation, including novelty, challenge, excitement and pleasure , which is likely to predict user perception of e-learning enjoyment. Therefore, we include self-enhancement-type values (achievement and power) as predictors of performance expectancy price value, and social influence and the value hedonism as a predictor of hedonic motivation.
Facilitating conditions was excluded from the final VETA model because it was not possible to create a reliable construct across the countries in this study. Values relating to self-transcendence and openness-to-change were excluded from the model, with the exception of hedonism. Firstly, values relating to self-transcendence relating primarily to an individual's views on nature and society A. Mehta, et al. Computers & Education 141 (2019) 103617 were not relevant to the individual-level adoption of technology. Secondly, values relating to stimulation and self-direction, while potentially relevant to learning were not reliable as constructs within the data collected in this study.
Our research question for this study is: Does the combination of Unified Theory of the Acceptance and Use of Technology (UTAUT2) and Schwartz's Theory of Human Values in a new model (Values Enhanced Technology Adoption (VETA)) explain learner intention to use e-learning in The Gambia and the UK?

Sampling
Two organisations participated in the study, one organisation in the UK, one organisation in The Gambia. Study participants were workers in each organisation who undertook the professional e-learning courses developed for this study. The e-learning course and survey were made available to all workers in each organisation for a duration of 3 months during 2016. Workers participated in the elearning and self-selected into the survey. Workers in the UK were involved in the collection of geological data, workers in The Gambia were involved in the collection of clinical trial data. Both UK workers and Gambian workers had a variety of educational and professional backgrounds. Ethics approval was obtained in each country. The details of each group sample are shown in Table 2.

Survey design
E-learning was created in Articulate Storyline 2 to provide training relevant to the work of the participants. Workers in the UK received training on computer use, workers in The Gambia received training on community engagement. Embedded within the elearning introductory screens was a three-part survey (deployed through www.onlinesurveys.ac.uk) where participants answered questions about demography, UTAUT2 (as shown in Section 12) and values. UTAUT2 adapted items (Venkatesh et al., 2012) (shown in Section 12) were scored on a 7-point Likert scale (7 = Strongly Agree, 6 = Agree, 5 = Slightly Agree, 4 = Neutral, 3 = Slightly Disagree, 2 = Disagree, 1 = Strongly Disagree) such that a high score indicated high perception of the latent variable. Values items were used directly from Schwartz's Portrait Values Questionnaire (PVQ5x) on a 6-point Likert scale (6 = Very much like me, 5 = Like me, 4 = Somewhat like me, 3 = A little like me, 2 = Not like me, 1 = Not like me at all) ) (shown in Section 12).

Data screening and software
Partial Least Squares (PLS) was used to test the model, using the software SmartPLS (version 3.2.6, SmartPLS GmbH) (Ringle, Wende, & Becker, 2015). Means and standard deviations were calculated in the Statistical Package for Social Sciences (SPSS version 22, IBM). Data were cleaned to remove partial responses and outliers using a z-score of 3 as a threshold for removal (Tripathy, 2013). This process yielded 160 useable responses from The Gambia, and 113 from the UK.

The sample
Demographic information for the two country groups is shown in Table 2. The UK sample was relatively balanced in terms of gender (56% male) compared to the Gambian group (77% male). Both groups were spread across the working population in terms of age, with the Gambian group reflecting a lower retirement age, with no participants over 60 years old. Reliability and validity data for the two country groups are shown in Table 3, Table 4, with the measurement model data shown in Table 5.

The measurement model
The measurement model was constructed using UTAUT2 as a base with values as predictors of adoption factors as theoretically proposed in the model development section above. Values were constructed as per their validated inventories Venkatesh et al., 2012). Items with outer loadings below 0.7 in both groups were considered for removal if doing so increased composite reliability (CR) or average variance explained (AVE) in the construct (Joseph F. Hair, Hult, Ringle, & Sarstedt, 2014;Sarstedt, Ringle, & Hair, 2017). Item loadings were above 0.6 in each country group with the exception of item COI1 in The Gambian group. Items removed from the final model are shown in section 12. Tables 3 and 4 show the convergent and discriminant validity data for the VETA model for each country group. Composite Reliability (CR) for constructs in each country group was above 0.7 (Dijkstra & Henseler, 2015;Hair, Ringle, & Sarstedt, 2011;Nunnally & Bernstein, 1994). Average Variance Extracted (AVE) was above the threshold of 0.5 for all constructs in each country group (Hair, Anderson, Tatham, & Black, 1998b). Divergent validity was established by the Fornell-Larcker criterion: latent variable correlations exceeded the square root of AVE for each construct (Fornell & Larcker, 1981) (Tables 3 and 4). Outer VIF values were below the threshold of 5, indicating collinearity was not of concern (Hair, Anderson, Tatham, & Black, 1998a;Pan & Jackson, 2008;Rogerson, 2001;Venkatesh et al., 2012). Inner VIF values were below 5, indicating that common method variance was not of concern (Kock, 2015). Therefore, the data in the measurement model was sufficient to proceed with analysis in the structural model.

The structural model
The VETA model (Fig. 2) explained similar variance (Adj. R 2 ) in behavioural intention to use e-learning in The Gambia (53%) and the UK (52%), comparable with the literature (Venkatesh et al., 2012). Variance explained was comparable across countries for performance expectancy (48% in The Gambia, 42% in the UK); and for hedonic motivation (5% in The Gambia, 4% in the UK). The variance in price value explained by the model differed in The Gambia (55%) and in the UK (4%), indicating a difference between groups in the perception of value from e-learning, and in social influence (18% in The Gambia, 4% in the UK) indicating a difference in perceptions of social context.
Influence on behavioural intention came from price value, performance expectancy and habit in both The Gambia and the UK; the predictors of performance expectancy were price value in both countries, with hedonic motivation, effort expectancy and social influence  A. Mehta, et al. Computers & Education 141 (2019) 103617 predicting performance expectancy in one country group only. Social influence and hedonic motivation had no direct predictive influence on behavioural intention in either country group. The strongest influence from values on the model came from the priority of achievement on price value, social influence and performance expectancy; priority of hedonism on hedonic motivation, and priority of tradition on social influence. The structural VETA model for both The Gambia and the UK is shown in Fig. 2. For The Gambia, Composite Reliability (CR) were between 0.77 and 0.93, and Average Variance Extracted (AVE) were all between 0.50 and 0.80 and greater than squared latent variable correlations indicating adequate convergent validity and internal consistency. Mean response and standard deviation (SD) is shown. Diagonal elements (bold) are the square root of AVE; off-diagonal elements are latent variable correlations. For discriminant validity, diagonal elements should be greater than off-diagonal elements.
For the UK, Composite Reliability (CR) were between 0.82 and 0.93, and Average Variance Extracted (AVE) were all between 0.55 and 0.88 and greater than squared latent variable correlations indicating adequate convergent validity and internal consistency. Mean response and standard deviation (SD) is shown. Diagonal elements (bold) are the square root of AVE; off-diagonal elements are latent variable correlations. For discriminant validity, diagonal elements should be greater than off-diagonal elements.
5.3.1. The UTAUT2 core of the VETA model Performance expectancy (PE) predicted behavioural intention (BI) in both country groups indicating that both Gambian and UK learners intended to use e-learning where it was perceived as useful in achieving their goals (Fig. 2, β[The Gambia] = 0.169, p = 0.042; β[UK] = 0.236, p = 0.006). Price value (PV) predicted behavioural intention (BI) in the both The Gambia and the UK (Fig. 2, β[The Gambia] = 0.360, p < 0.001; β[UK] = 0.258, p = 0.013), such that learners intended to use e-learning where it was perceived as worth the effort that they perceived they needed to contribute towards use. Habit (HAB) predicted behavioural intention (BI) in both The Gambia and the UK (Fig. 2, β[The Gambia] = 0.261, p = 0.005; β[UK] = 0.216, p = 0.005), indicating that prior experience with online learning was an important factor in determining learner behavioural intention to use e-learning irrespective of context. Price value (PV) predicted performance expectancy (PE) (Fig. 2, β[The Gambia] = 0.370, p < 0.001; β[UK] = 0.300, p = 0.004) such that learners perceived e-learning as useful if the benefits outweighed the costs.
Effort expectancy (EE) directly predicted behavioural intention (BI) for UK learners (Fig. 2, β[UK, EE-BI] = 0.208, p = 0.005), supporting research in developed contexts, but acted via performance expectancy (PE) in The Gambia (Fig. 2, β[The Gambia, EE-PE] = 0.252, p = 0.006), indicating that while UK learners intended to use e-learning where it was easy to use, Gambian learners formed a perception of e-learning usefulness which included ease of use amongst other factors. Hedonic motivation (HM) had no direct relationship with behavioural intention (BI) such that participants did not intend to use e-learning solely because they perceived it to be interesting or enjoyable (Fig. 2,

Integrating values into the UTAUT2 model
There was no influence of conformity, tradition or security on performance expectancy in either country group. The value of tradition had a significant influence on social influence (TR-SI) in the Gambian group (Fig. 2, β = 0.224, p = 0.003), with the path of conformity on social influence approaching significance at the p < 0.05 level in the Gambian group (Fig. 2, β = 0.157, p = 0.053). There was an influence of achievement (AC) in the VETA model: the path (AC-PV) was important in The Gambia (β = 0.239, p = 0.005) but not in the UK. An individual's priority on the value of achievement predicted their perception of social influence (AC-SI) in The Gambia (β = 0.263, p = 0.013), but not in the UK. The influence of achievement on performance expectancy was negative in the UK (β = −0.178, p = 0.026), and non-significant in The Gambia. An individual's priority for the value of hedonism predicted learner's perception of hedonic motivation with a positive and significant path in the Gambian group with the path in the UK group also approaching significance at the p < 0.05 level (Fig. 2, β[The Gambia] = 0.273, p = 0.002; β[UK] = 0.247, p = 0.067).

Discussion
In this study, we proposed the link between Schwartz's theory of human values and the Unified Theory of Acceptance and Use of Technology (UTAUT2) to develop the Value-Enhanced Technology Adoption (VETA) model, an extended UTAUT2 model, and validated the VETA model against the context of professional e-learning in The Gambia and the UK. The study contributes to the literature through the findings that learners who prioritise self-enhancement values and therefore prioritise their own achievement perceive e-learning as worthwhile and that their social environment endorses e-learning use. The study contributes to the literature by finding no links between values associated with security, conformity and power and the adoption model for e-learning. The VETA model explained learner intention to use e-learning in The Gambia and the UK to an extent comparable with the UTAUT and UTAUT2 models. The values constructs in the VETA model explained variance in adoption variables: performance expectancy, hedonic motivation, price value and social influence. The VETA model, therefore, contributes to the literature by explaining learner intention as well as learner perception of adoption factors in The Gambia and the UK.

UTAUT2 was partially validated
This study supports the magnitude and direction of the direct relationships of: performance expectancy; price value; and habit on behavioural intention. Performance expectancy predicted behavioural intention to use e-learning, in support of findings from the primary technology acceptance literature (Davis, 1989;Davis et al., 1989;Venkatesh et al., 2003Venkatesh et al., , 2012 and from previous e-learning studies (Ching-Ter, Hajiyev, & Su, 2017). Learners across The Gambia and the UK intended to use e-learning because they perceived that elearning was useful in achieving their goals. The effect of performance expectancy was smaller than expected but dominant when compared with the effect of the other UTAUT constructs, effort expectancy or social influence, confirming findings from meta-analysis (Dwivedi et al., 2011;Khechine, Lakhal, & Ndjambou, 2016;Taiwo & Downe, 2013). Price value was the strongest direct predictor of behavioural intention, supporting the primary UTAUT2 literature in direction and significance (Venkatesh et al., 2012). Learners intended to use e-learning based on an evaluation of the non-financial costs, such as time and effort taken to use e-learning around work priorities, and the benefits, such as the increased competence gained from e-learning. Habit predicted behavioural intention to use e-learning supporting part of the literature that has found an effect of habit (Alawan, Dwivedi, & Williams, 2013;Venkatesh et al., 2012;Yuan, Ma, Kanthawala, & Peng, 2015). Learners intended to use e-learning where they had established routines that involved online methods of learning.
This study contributes to theory development by adding second-order relationships in the UTAUT model, bridging the gap between e-learning adoption research (Abdullah & Ward, 2016) and the technology-agnostic information systems literature (Venkatesh et al., 2012). Performance expectancy mediated relationships between social influence, price value and hedonic motivation on behavioural intention to use e-learning, supporting the dominant place of perceived usefulness in the adoption literature (Dwivedi et al., 2011;Šumak et al., 2011;Taiwo & Downe, 2013). The direct effect of social influence on learner intention was non-significant supporting prior literature (Teo, 2011;Terzis & Economides, 2011) and meta-analysis (Taiwo & Downe, 2013). In the UK, there was an indirect relationship of social influence on learner intention via performance expectancy implying the influence was informational in nature, rather than normative, as learners gather information on usefulness from referent others in their environment. Hedonic motivation was similarly only significant in the UK, indicating a potential difference in how enjoyment in workplace e-learning was perceived as effectual in helping to achieve knowledge outcomes by the two cohorts.
The influence of effort expectancy differed between the UK and The Gambia. Effort expectancy only directly predicted behavioural intention to use e-learning for the UK group. This finding adds to the speculation that the place of the construct of effort expectancy as a technology-agnostic, context-agnostic direct predictor of intention in the UTAUT paradigm is uncertain (Baptista & Oliveira, 2016). UK learners intended to use e-learning because of ease of use directly, without the mediating effect of performance expectancy, supporting the literature from developed nation contexts (Tarhini, Hone, & Liu, 2014). The influence of effort expectancy differed by context, acting directly on intention in the UK, and via performance expectancy in The Gambia, reinforcing prior claims that the literature corpus generated in developed countries cannot be readily applied in the context of sub-Saharan Africa without additional research (Brown, 2002;Musa, 2006).

Self-enhancement values predicted the adoption model
The finding that learners who prioritised achievement perceived e-learning as worth the cost, was supported in the Gambian group. The construct of price value indicates a positive balance of cost and worth perceived by the learner, leading to three possible explanations for this finding, either: UK workers did not link e-learning with achievement; UK workers did not place worth on the perceived achievement that was linked with completing the e-learning course; or did not perceive that the worth assigned to e-learning was proportional to the cost incurred. Considering that to monetary cost to participants was zero, cost was theorised a priori to be incurred in other ways, such as time management or the cost of internet access. There is some indication in the literature that participants in sub-Saharan Africa have difficulties accessing infrastructure such as power, networks and devices (Folorunso, Ogunseye, & Sharma, 2006;Tarus, Gichoya, & Muumbo, 2015;Touray, Salminen, & Mursu, 2013). Therefore, for Gambian learners, travelling to a place that has appropriate network, power and device infrastructure or the cost of providing such infrastructure for themselves may be perceived costs.
The argument that cost/benefit proportionality explains the negative link between achievement and price value for UK workers is unlikely to be the explanation because the costs for UK workers were small. The explanation of this finding might, therefore, be linked to the worth placed by a learner on the achievement from e-learning. In the UK workplace e-learning is relatively abundant and might not precede career advancement; in the Gambian group, e-learning was relatively scarce, and further courses might link to career advancement (Njenga & Fourie, 2010).
The finding that learners who prioritised achievement perceived a similar sentiment in their reference group was only supported in the Gambian group. This finding indicates that the group perception of e-learning is linked to achievement, and that members of the group perceive e-learning as linked to a sense of achievement. This finding supports the literature that argues that African societies are more collectivist than the UK, and therefore place more emphasis on group influence (Hofstede et al., 2010).
The finding that learners who prioritise pleasure in their lives find e-learning enjoyable may be more complex than a trivial link between value and perception. With the literature emphasis on hedonic motivation as perception that has a target (a technology) (Brown & Venkatesh, 2005), and the investment priority in educational technology to make engaging and interesting learning products through creative instructional design, the finding that a learner's prior interest in engagement with educational technology determines intrinsic motivation indicates that producing engaging products may be less important than previously thought.
The lack of influence of conformity and security on the model were unexpected. The theoretical link between the priority that A. Mehta, et al. Computers & Education 141 (2019) 103617 individuals place on maintaining a status quo of subjective norms and the influence they feel from their reference group was not apparent in the data . Potentially, a loose mandate for professional e-learning removes the link to conformity and job security: if learners perceive e-learning in their workplace as encouraged but not mandatory, and there is no explicit reward or punishment, then ideas such as conformity and security may not be important. Linking with achievement values, if the achievement accrued through e-learning has no tangible benefit associated with it in terms of job security or career structure, then conformity and security would not be important in determining e-learning adoption. The link between these values and the adoption model should be explored further in future research.

Limitations
Due to the relatively small sample collected, it was not possible to run moderator analysis, so gender and age effects were not included. A larger study, incorporating different countries, organisations or different technologies, might provide the additional experimental power and data stability to investigate moderator effects, as well as the additional values in the self-transcendence and openness-to-change values dimensions to determine their influence on adoption. Similarly, it was not possible to develop a more complex model, with constructs for facilitating conditions, self-direction and stimulation proving unreliable with the small sample of data collected. Additional data would add to the strength of the findings and the explanatory power of the model.
It was not possible to conduct the study on two organisations with similar worker profiles: UK workers have an infrastructure which supports technology, and as such, workers are comfortable with computing devices; Gambian workers have an infrastructure that is susceptible to power and network instability, and lack of device availability. It was not possible to conduct random sampling due to the small populations of workers in the target organisations, instead, all workers were invited to participate, resulting in selfselection, which is a limitation of this study.

Implications for practice
The importance of achievement has important implications for practitioners: e-learning that is perceived as yielding low achievement is unlikely to be linked to an intention to adopt the e-learning. Linking this to prior literature that indicates prior experience of low-value e-learning is likely to yield poor learner perception of e-learning, since users may not always separate elearning into content, pedagogy, and the complex network of hard and soft technology, practitioners should ensure that the content, design, pedagogy, implementation, and technology are all perceived as high value to learners if a course is to be adopted.
Practitioners implementing e-learning in workplace contexts should also be mindful of the link between perceived cost and usefulness in achievement. Practical considerations for training managers should include providing adequate time and resources to complete e-learning, and providing appropriate reward following e-learning, such as a certificate of achievement and follow-on courses for further development.
Organisations can apply e-learning in similar manners in different cultures but must be mindful that what works in one might not work in another. It is therefore important to investigate what learners in each organisation deem as expected outcomes, costs and who key players in the social environment might be, before investing in an e-learning project.

Future research
To explain similarities and differences, it is important to expand the differing perceptions of UTAUT2 variables by context, therefore a qualitative study would be beneficial to deconstruct these variables. With the VETA model developed and validated in this study, further work is required to generalise findings to other contexts, to determine the breadth of applicability of the model.
Although this study uses e-learning as a vehicle to test the model, the nature of human values being scenario-agnostic, and UTAUT being technology-agnostic, mean that the VETA model could be applied to other scenarios. Although links between the dimensions of self-transcendence and individual-level adoption were not included in the VETA model for e-learning adoption, they should be considered when applying the model to technologies that have societal impact.

Conclusion
Schwartz's theory of human values was used to extend the UTAUT2, to develop the Values-Enhanced Technology Adoption (VETA) model, which was validated in the context of e-learning adoption in The Gambia and the UK. This exploratory study demonstrated that the VETA model partially applied in both contexts, with behavioural intention predicted by performance expectancy, price value and habit, and the value of achievement being an important predictor of adoption factors.

Funding and approvals
Funding for this study was provided through the e-learning project budget for developing a fieldwork curriculum at the Medical Research Council, UK. Science and ethics approvals were received as follows: University of Leeds, UK (AREA 14-018); The Gambia (SCC-1403).

Conflicts of interest
The authors declare no conflicts of interest.