Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) REID (Research and Evaluation in Education), 7(2), 2021, 132-144 Available online at: http://journal.uny.ac.id/index.php/reid The effect of external factors moderated by digital literacy on the actual use of e-learning during the Covid-19 pandemic in Islamic universities in Indonesia Sumin1*; Kahirol Mohd Salleh2; Nurdin3 1Institut Agama Islam Negeri Pontianak, Indonesia 2Universiti Tun Hussein Onn Malaysia, Malaysia 3Sekolah Menengah Kejuruan (SMK) Negeri Taman Fajar Aceh Timur, Indonesia *Corresponding Author. E-mail: amien.ptk@gmail.com INTRODUCTION Since the initial appearance of a new variant of the Coronavirus in Wuhan-China at the end of 2019, which was later named Sars-Cov 19, the pandemic has infected nearly 200 million peo- ple. Based on WHO data (03/08/2021), 223 countries have been affected by Covid-19, with 198.778.175 confirmed cases, 4.235.559 deaths, a total of 3.886.112.928 doses of vaccine have been given (UNESCO, n.d.). The Covid-19 pandemic, both directly and indirectly, has had a wide impact on human life. In addition to causing health impacts, Covid-19 has also harmed almost all sectors. The signifi- cant impact felt by almost all countries confirmed by Covid-19 is the impact on the economic, social, political, and educational sectors. In the economic sector, the distribution of goods and services can only be carried out with strict restrictions or health protocols, as well as culinary businesses and the Micro, Small and Medium Enterprises (MSME) sector, having to close their businesses because of the govern- ment's policy of restrictions to tackle the transmission of Covid-19, some are even forced to quit or go bankrupt because few customers come to make transactions, so they are unable to pay off ARTICLE INFO ABSTRACT Article History Submitted: 27 October 2021 Revised: 9 December 2021 Accepted: 9 December 2021 Keywords external factors; actual use; digital literacy; e-learning Scan Me: The future of education is increasingly worrying due to the impact of the Covid-19 pandemic since the end of 2019. Restrictions on community activities and campus closures have forced university administrators to use e-learning. On the other hand, online learning has encountered many obstacles. Barriers to the use of e-learning are thought to stem from external problems (online facilities and infrastructure) or edu- cators and students (internal factors), such as lack of literacy, low absorption, level of understanding, and other non-technical factors. This study aims to examine further the influence of external factors and digital literacy and the moderating effect of digital literacy with external factors (System Design, User Friendly, Devices, Internet, Elec- tricity) on the actual use of campus e-learning at Islamic universities. This study found that: External factor variables have a significant positive effect on the actual use of e- learning. The digital literacy variable has a significant positive effect on the actual use of e-learning. The digital literacy variable weakens the influence of external factors on the actual use of e-learning at Islamic universities in Indonesia. This is an open access article under the CC-BY-SA license. How to cite: Sumin, S., Salleh, K., & Nurdin, N. (2021). The effect of external factors moderated by digital literacy on the actual use of e-learning during the Covid-19 pandemic in Islamic universities in Indonesia. REID (Research and Evaluation in Education), 7(2), 132-144. doi:https://doi.org/10.21831/reid.v7i2.44794 https://creativecommons.org/licenses/by-sa/4.0/ https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 133 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) their investment loans. In the social sector, the hustle and bustle of night entertainment, social and religious associations, and communities were forced to stop carrying out activities to prevent the transmission of Covid-19. In the political and government sectors, Covid-19 has forced governments in affected countries to make policies to prevent and control Covid-19. Even the government should cut and do refocusing the budget to provide social assistance, accelerate the absorption of vaccination, and breaking the chain of transmission of Covid-19. The impact of the Covid-19 pandemic on the Education sector can be seen from the offi- cial report of the United Nations Educational, Scientific and Cultural Organization (UNESCO), it was reported that: One year after the COVID-19 pandemic, almost half of the world's students are still affected by partial or complete school closures, and more than 100 million additional chil-dren will fall below the minimum reading proficiency level as a result of the health crisis. Priori-tizing the restoration of education is critical to avoiding a generational disaster as highlighted at the ministerial summit in March 2021. (UNESCO, n.d.). UNESCO supports countries in their efforts to reduce the impact of school closures, address learning loss and adapt education systems, especially for vulnerable and disadvantaged communities. To mobilize and support sustainable learning, UNESCO has formed the Global Education Coalition which currently has 160 members working on three main themes: Gender, connectivity, and teachers (UNESCO, n.d.). The future of education is very worrying and is at stake due to the direct and indirect im- pacts of the current Covid-19 pandemic. Learning conducted through online media (e-learning or social media), has encountered many obstacles. Barriers to the use of e-learning can come from external problems (online facilities and infrastructure) as well as from factors originating from within the educators and students (internal factors) such as lack of literacy, low absorption, level of understanding, and other non-technical factors. Referring to the results of research by Putro et al. (2020), the five countries that are the subject of this study (Indonesia, the Philippines, Nigeria, Finland, and Germany), are divided into groups of poor, developing and developed countries. Problems of online learning during the Covid-19 pandemic found in research Putro et al. (2020) can be classified into two groups, namely technical and non-technical problems. Technical problems include the issue of inadequate internet network availability, unsupported electricity network, limited availability of supporting equipment (facilities) both in terms of educators and students, and school access to software that supports online learning is not evenly distributed. The non-technical factors include the problem of the level of understanding of the material taught online, the many and piling up of assignments submitted online, the lack of equitable mas- tery of information technology from educators and students, the uneven financial ability of stu- dents, the online learning system is easy to cause problems. boredom in students, disruption of household work (because learning must be from their respective homes), students' difficulties in adapting to online learning systems, online learning systems reduce verbal communication be- tween educators and students, disruption of the academic calendar (due to policy changes), rou- tine training and seminar activities are eliminated, there is a learning gap, education budget cuts, interference from family, friends and the environment (because learning is carried out from home), lack of response and good coordination in overcoming obstacles in online learning, I said. the lack of distance learning experiences, the high cost of the internet, the poor condition of edu- cation in the pre-pandemic period, the lack of parental support and understanding of online learning, rapid changes that result in insufficient time to make appropriate learning designs, the gap between students who have high ability and good mentality in learning with students with the opposite nature, online group assignments are less effective, fatigue, there is a view that technol- ogy in the world of education is not good for children. The non-technical ability of educators and students in using online learning systems as de- scribed in a study by Putro et al. (2020) cannot be separated from the digital literacy of educators https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 134 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) and students, this is following the results of the study Widana (2020, p. 8) who found that “…the digital literacy factor is one of the variables that significantly affect the ability of teachers to de- velop HOTS-based assessments…” Widana's research results are in line with a research by Noh (2017) who found that “…bit literacy most influences information usage behavior, followed by virtual community literacy and technical literacy… Bit literacy is related to the ability to use in- formation including information retrieval, information acuity, information editing, information processing, and information utilization…” A research presented by Jan (2017, p. 31) found that “… digital literacy (DL), tablet and smartphone use, previous training in computer use and frequency of computer use significantly influence students' attitudes toward ICT use”. On the other hand, the results of Jang et al. (2020) found that digital literacy had no direct significant effect on the intention to use. On the other hand, the research conducted by Jang et al. (2020) found that “Digital Literacy did not have a direct significant effect on the intention to use learning technology in Finland...” The researchers chose the state Islamic universities (Perguruan Tinggi Keagamaan Islam Negeri or PTKIN) as the research locus, considering that the majority of students and lecturers at PTKIN were graduates of traditional Islamic boarding schools, considering the results of a re- search by Azzahra and Amanta (2021) that most Islamic boarding schools in Indonesia use tradi- tional learning systems so that it is difficult to obtain digital literacy and use digital technology. There is no complete information regarding how many Islamic boarding schools are equipped with facilities such as the internet and computers. Thus, based on the research gap and the results of previous studies, this paper aims to further examine the influence of external factors moder- ated by digital literacy skills on the use of e-learning at state Islamic universities in Indonesia. External Factors Given the breadth of the definition of external factors, the researchers narrowed the mean- ing of external factors only to the context or scope of learning information technology. Limiting the meaning of external factors is intended to facilitate the identification of variables and their indicators. The external factors referred to in this study are related to factors originating from outside the students themselves that can affect their ability to utilize information technology. External variables can be defined as factors outside the users of information technology, such as; information system design, easy to understand or easy to learn, availability of supporting devices (Smartphone/Computer), adequate Internet network, adequate electricity network. Ac- cording to Davis et al. (1989, p. 985), perceived usefulness can be influenced by several external factors, educational programs are designed to "capture" potential users to use information sys- tems to increase user productivity. Still according to Davis et al. (1989, p. 985), learning based on the concept of feedback between educators and students is another type of external variable that can affect beliefs in the use of information technology. Digital Literacy According to Krumsvik (2008) in Liu et al. (2020), “Computer-based literacy, media liter- acy, digital literacy, and digital competence are concepts that focus on the need to use technology in the digital era”. Meanwhile, according to Kaeophanuek et al., (2019) digital literacy is: …a set of competencies possessed by an individual to apply digital tools well in the digital era, easily accessing, applying, evaluating, analyzing and synthesizing data, as well as creating new knowledge. With that, students will be able to communicate and present content through various digital technologies. A good level of digital literacy will make it easier for students to achieve their goals. If literacy is defined as a person's ability to read, interpret written sources of knowledge in a social group, then academic literacy is the ability to read, interpret and produce information in a digital format that is valued in academia (Kaeophanuek et al., 2019, p. 24). https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 135 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) According to Ferrari; Kaeophanuek, et al.; and Owen, et al. in Kaeophanuek et al. (2019, p. 24), the indicators that form digital literacy variables can be identified as follows. (a) Information skills: The basis for information management, techniques, and various strategies involving digital information management, which includes the process of identifying problems, determining the search topics, methods, and strategies for accessing, analyzing, synthesizing content, systematiz- ing, evaluating, interpreting, and applying information used in doing or solving problems cor- rectly. (b) Use of digital tools: Skills and competencies in learning how to use a wide variety of software and applied digital tools to successfully accommodate everyday life. It also relates to the ability to maintain, manage use and troubleshoot basic computers, as well as the ability to com- municate, systematically manage either personal or network data, comply with ethical norms, and utilize technology for effective teamwork. (c) Digital transformation: Skills in consolidating infor- mation to create, improve, design, and producing content and products, and presenting informa- tion in the form of new information, creating new knowledge and new digital innovations under collaborative learning. Learners can reflect on their thoughts to improve their work and publish it following copyright laws. Actual System Use Actual use (actual system use) is a real condition of the application of information systems/ information technology measured in units of time or frequency of technology use, users will feel satisfied while using information system services or information technology if they believe that the system has been used, can increase productivity work that is reflected in the real conditions of use (Davis et al., 1989, p. 987; Venkatesh & Davis, 2000, p. 204). Measurement of actual usage (actual system usage) can be evaluated in units of time, how often users use information technol- ogy service systems in terms of duration of usage time. Actual technology use is measured by the amount of accumulated time spent interacting with technology and the number of times using the technology (Davis et al., 1989, p. 987; Venkatesh & Davis, 2000, p. 204). According to Li and Lalani in Hermawan (2021), “E-learning can be found from a variety of existing learning media, ranging from language applications, video conferencing tools, virtual tutoring, online learning software, Moodle, and many more”. Based on the opinions of experts that have been put forward, an online learning system (e- learning) is a set of software designed as an online-based learning medium (using an internet net- work) that can be accessed via desktop computers and mobile devices (smartphones) containing learning plans, learning process, and learning evaluation. Learning plans can be designed and in- corporated into e-learning based on the curriculum content that has been set in the learning cur- riculum document. The learning process through e-learning includes delivering material in various forms, such as textbooks, journals, presentation slides, audio videos, or learning website addresses, recording the attendance of teachers and students, and discussion rooms. E-learning applications at certain universities or schools have been designed following the requirements of the quality standards of education with special criteria and conditions, which are different from social media or e-learning based on content management systems offered by software develop- ment companies such as Google. classroom, moodle, and others. According to Davis et al. (1989), External variables have no direct effect on attitudes and behavior in using technology, but the technology acceptance model proposed by Davis et al. (1989) found that there is a procedure that bridges between external variables and attitudes in using technology, it is triggered by individual differences related to one's personality and charac- teristics, which are related to one's self-confidence and belief. A research by Jan (2017, p. 31) found that “… digital literacy (DL), tablet and smartphone use, previous training in computer use and frequency of computer use significantly influence stu- dents' attitudes toward ICT use”. On the other hand, the results of Jang et al. (2020) found that digital literacy had no direct significant effect on the intention to use. https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 136 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) Figure 1. Thinking Framework The researchers have not found any previous studies that place digital literacy as a moder- ating influence between external factors on the actual use of information and computer technol- ogy (ICT), therefore the researchers intend to develop a model that has been found previously by experts, by placing digital literacy as a moderating variable of the influence between external fac- tors on the actual use of e-learning. Based on the theories and results of previous research, this research model can be described as in Figure 1. METHOD This study uses a correlational research design and a quantitative approach. The use of the correlational method is to determine the coefficient of the effect of the predictor variables on the response variable (Creswell, 2012, p. 45). The correlational research design is used to analyze the direct influence of external variables and digital literacy variables on the use of e-learning, as well as to measure the impact of digital literacy variables which are hypothesized to strengthen or weaken the influence of external variables on the use of e-learning. Based on the theory that underlies the relationship between external factors and digital lit- eracy on the use of learning information technology (e-learning), the researchers formulate the following hypothesis. H1: There is a significant direct effect of external factors on the actual use of e-learning at Islamic universities in Indonesia; H2a: There is a direct positive and significant in- fluence of digital literacy on the actual use of e-learning at Islamic universities in Indonesia; H2b: Digital literacy variables can strengthen the influence of external factors on the actual use of e- learning at Islamic universities in Indonesia. To answer the research questions, the researchers used one of the units of multivariate statistical analysis, namely; structural equation modeling (SEM) with partial least squares (PLS) approach. Herman Word introduced SEM-PLS to model latent variables which he called "soft modeling". The term refers to the flexibility of using SEM-PLS which does not require many as- sumptions and does not have to be based on a strong theory. SEM-PLS can be used for theory confirmation but can also be used to develop models (Vinzi et al., 2010, p. 2). The analysis of Structural Equation Modeling with the Partial Least Square (PLS) approach consists of two stages: Evaluation of the measurement model in the SEM-PLS analysis is called the Outer Model, and the evaluation of the structural model, in terms of SEM-PLS, is called the Inner Model. Model estimation in SEM-PLS is carried out in two stages, namely: first; conducting an assessment of the measurement model (outer model), second; conducting an assessment of the structural model (Inner model). The measurement model is defined through the equations in Formula (1) and Formula (2), in which = Factor load from indicator to latent variable, = Re- External Factors (EF) Digital Literacy (DL) Actual Usage (AU) DL*EF H1 H2a H2b https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 137 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) sidual/error on the exogenous latent variable indicator, = Residual/error on the endogenous latent variable indicator, = Exogenous latent variable indicators, = Endogenous latent variable indicators. …………………….. (1) ….………...………… (2) The quality of the measurement model can be assessed from several measuring instruments related to the instrument items’ validity and reliability (Garson, 2016; Hair et al., 2014; Lohmöller, 1989), that is: (1) Convergent Validity; an instrument item can be designated as a valid measure ment if it has a loading factor value of ≥ 0.7, (2) Discriminant Validity; a variable measurement item can be designated as a good measurer, if it has the item only significant in measuring the la- tent construct/variable in its indicator block, and should not be significant in other latent variable indicator blocks, and (3) Composite Reliability and Internal Consistency; instrument items meet composite reliability if they have a composite reliability coefficient (CR) > 0.7, and are considered to have good/ideal internal consistency if they have Cronbach's Alpha coefficients > 0.7. The structural model (inner model) can be defined through the following mathematical equations in Formula (3), where Endogenous latent variable, Exogenous latent variable, Parameter coefficient (factor loading) of exogenous latent variable to endogenous latent variable, and Residual/error of inner model. The inner model can be assessed from several statistical measures as follows: (a) Model fit (fit index, coefficient of determination, and effect size), and (b) Hypothesis Testing (direct and indirect effect, total effect). …………. (3) The sampling technique used in this paper is Simple Random Sampling. Sample Size is cal- culated using the sample formula from Lemeshow et al. (1990), as shown in Formula (4), where Z = Standard normal table value (if the confidence level is 95%, then the Z value = 1.96), = The estimated proportion of the attributes that are in it, is assumed to be 0.5, and e = The abso- lute precision required. ……………….. (4) If set value; confidence interval (1-α) = 95% (α=5%), the standard normal distribution table value is 1.96. Anticipated population proportion = 0.80 assuming the proportion of stu- dents using e-learning at PTKIN is 80%. The expected absolute precision was set at =0.03999 (close to the 5% significance level). Relative precision (ε), equal to = 0.0499875 (close to 5% sig- nificance level). The population size is 760.619. Then, the sample size can be calculated based on Formula (4), as follows. The ideal number of samples in this study were 385 respondents, taken randomly from a student population of 760.619 from 59 state Islamic universities in Indonesia. Primary data in this https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 138 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) study were obtained through a questionnaire. Questionnaires were distributed randomly to the respondents who were the target sample (students at state Islamic universities in Indonesia). The data collection instrument used was a questionnaire. This questionnaire was designed using the Symantec Differential Scaling scale. Symantec Differential Scaling is an attitude scale arranged in a continuum line, with very positive answers on the right and very negative answers on the left. The Symantec differential uses seven response options for the statement of each instrument item, with categories for positive questions graded from "Strongly Disagree" with an item score = 1 to "Strongly Agree" with score of = 7, on the contrary for negative questions graded from "Strongly Agree" with item score = 1 to “Strongly Disagree” with score = 7 (Rosenberg & Navarro, 2018). Convergent validity checking on the reflective measurement model can be seen from the standardized loadings factor value which shows the correlation between indicator scores and la- tent variables, as in Table 1. Calculations using WarpPLS 7.0 software on 14 items of external factor variable indicator items and 23 digital literacy variable indicator items and seven item indi- cator variables for the actual use of e-learning, obtained the following results; items from the la- tent variables of external factors x1.7, x1.8, x1.9, x1.10, x1.11, x1.12, and x1.13 have standardized loadings factor values < 0.7 so that seven items must be discarded because it is not valid in measuring the latent variable of external factors. Table 1. Variables and Indicators of Measurement of Research Variables No. Variable Indicators Measuring Scale 1. External Factors (Davis et al., 1989) ▪ E-learning design ▪ Easy to Understand (user-friendly) ▪ Online Device Availability (Smartphone/ Computer) ▪ Adequate Internet Network Availability ▪ Availability of adequate electricity grid Ordinal 2. Digital Literacy (Kaeophanuek et al., 2019) ▪ Skills in managing information ▪ Skills in using digital equipment ▪ Digital transformation Ordinal 3. Actual Usage of e-learning (Davis et al., 1989; Venkatesh & Davis, 2000) ▪ Usage time duration ▪ Frequency of use Ordinal Table 2. Convergent Validity Checking Item EF DL AU Indicator Type Annotation x1.1 0.777 0.192 0.052 Reflective Valid x1.2 0.771 -0.03 0.065 Reflective Valid x1.3 0.762 -0.06 0.122 Reflective Valid x1.4 0.798 -0.04 -0.116 Reflective Valid x1.5 0.739 -0.04 -0.176 Reflective Valid x1.6 0.751 -0.02 0.052 Reflective Valid x2.3 0.705 -0.009 Reflective Valid x2.8 0.781 -0.002 Reflective Valid x2.11 0.781 0.013 Reflective Valid x2.12 0.804 -0.002 Reflective Valid x2.14 0.735 -0.113 Reflective Valid x2.17 0.83 0.038 Reflective Valid x2.19 0.719 -0.041 Reflective Valid x2.20 0.766 0.039 Reflective Valid x2.21 0.754 0.068 Reflective Valid y1.1 0.847 Reflective Valid y1.2 0.859 Reflective Valid y1.3 0.743 Reflective Valid y1.4 0.853 Reflective Valid y1.5 0.805 Reflective Valid https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 139 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) Furthermore, of the 23 items measuring the digital literacy latent variable indicator, 13 items have a standardized loading factor value of < 0.7, namely; items x2.1, x2.2, x2.4, x2.5, x2.6, x2.7, x2.9, x2.10, x2.13, x2.15, x2.15, x2.16, x2.18, x2.22 and x2.23, these items are not valid in measuring digital literacy latent variables and must be excluded, because they are not valid in measuring digital literacy variables. In the variable of actual use of e-learning, there are two items of the question- naire that have a standardized loading factor value of < 0.7 out of seven of the questionnaire items used to measure the actual use of e-learning, namely; y1.6 and y1.7, the items must also be ex- cluded from the measurement, because it is not valid in measuring the actual use of e-learning variables. The results of measuring variables after removing instruments that do not meet the cri- teria for convergent validity in this study are presented in Table 2. A measuring instrument is said to have good validity, not only judged by its ability to meas- ure the variables in the contract to be measured but also to be significantly different from meas- urements in other construct indicator blocks. To find out how far the instrument items differ from one indicator block to another, it can be seen from the results of discriminant validity test- ing. The results of the discriminant validity test in SEM-PLS using WarpPLS software in this pa- per are presented in Table 3. The results of the discriminant validity test in Table 3 show the average value of the variant extract (AVE) is below the root value of the average variant extract, so it can be concluded that the measurement item on the external factor latent variable is only valid for measuring external factor variables, thus also the items used in the digital literacy variable and the actual use of e- learning. The results of instrument reliability testing using composites and internal consistency (Alpha Cronbach) in this paper are presented in Table 4. Table 3. Discriminant Validity Checking Variable AVE Square of AVE Annotation EF 0.588 0.766812 Valid DL 0.585 0.764853 Valid AU 0.677 0.8228 Valid Source: Primary Data, processed with SEM-PLS (WarpPLS 7.0), 2021. Table 4. Composite Reliability Checking Variable Composite Reliability Coefficients Internal Consistency (Cronbach's Alpha Coefficients) Annotation EF 0.895 0.859 Reliable DL 0.927 0.911 Reliable AU 0.913 0.88 Reliable Table 5. Model Accuracy Criteria No. Model fit and quality indices Statistics Criteria (Kock, 2019) Annotation 1. Average path coefficient (APC) 0.296 P<0.05 Accepted 2. Average R-squared (ARS) 0.52 P<0.05 Accepted 3. Average adjusted R-squared (AARS) 0.517 P<0.05 Accepted 4. Average block VIF (AVIF) 1.2 Accepted ≤ 5, Ideal ≤ 3.3 Ideal 5. Average full collinearity VIF (AFVIF) 1.621 Accepted ≤ 5, Ideal ≤ 3.3 Ideal 6. Tenenhaus GoF (GoF) 0.609 Small ≥0.1, Medium ≥ 0.25, Large ≥ 0.36 Large 7. Sympson's paradox ratio (SPR) 1 Accepted ≥ 0.7, Ideal = 1 Ideal 8. R-squared contribution ratio (RSCR) 1 Accepted ≥ 0.9, Ideal = 1 Ideal 9. Statistical suppression ratio (SSR) 1 Accepted ≥ 0.7 Ideal 10 Nonlinear bivariate causality direction ratio (NLBCDR) 1 Accepted ≥ 0.7 Ideal https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 140 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) The results of composite reliability test and Cronbach's Alpha internal consistency show that all variables have a measurement coefficient value > 0.7. It indicates that all measurement items of external factor variables, digital literacy, and actual use of e-learning are reliable and con- sistent in measuring their respective latent variables. The suitability of the model with the theory that underlies the relationship between variables can be seen from the model's accuracy index. It can be interpreted by how accurately the research model confirms the results of previous studies. If the model accuracy index is met (ideal and significant), it means; The model developed is stated to be appropriate and has succeeded in confirming the results of previous studies. The SEM-PLS using WarpPLS 7.0 application, presents ten types of model accuracy indexes as seen in Table 5. Based on the ideal criteria and the accuracy index value of the model obtained, it can be concluded that the model developed in this study is appropriate or does not violate the results of previous studies. SEM-PLS in principle is a development of regression analysis, for that, it is re- quired that the indicator block must be free of multicollinearity which is marked by the value of Variance Inflation Factors (VIF)<3 The value of variance in this paper can be seen in Table 6. Taking into account the value of the latent variable indicator block VIF in Table 6, it can be seen that all latent variables, including the moderating variable between digital literacy and external factors, have a VIF value <3.3, which can be interpreted that all indicators are multi- collinearity free. The size of the effect (effect size) is a measure of the meaning of the results of research at a practical level. The effect size criteria are; if the value of effect size (f) = 0.1 means it has a small effect size, f = 0.25 has a medium effect size, and f = 0.4 has a large effect size (Kock, 2019). The effect size value of this paper can be seen in Table 7. The effect size value in Table 7 can be interpreted that, the size of the influence of external factor variables on the actual use of e-learning is moderate, the influence of digital literacy on the actual use of e-learning is moderate and the interaction of digital literacy with external factor vari- ables on the actual use of e-learning relatively small. The coefficient of determination shows the magnitude of the contribution of the exogenous latent variable to the endogenous latent variable. The results of the analysis show that the R-Square value of 0.52, it can be interpreted that the variation of the actual use of e-learning variables can be explained by external factors, digital liter- acy and moderation between digital literacy and external factors is 52.00% while the remaining 48.00% is explained by other variables not included in this study. Hypothesis testing using SEM-PLS can be assessed from the path coefficient. This meas- urement is obtained through the estimation of the parameters of the research model. The results of the SEM-PLS parameter testing using the WarpPLS 7.0 software are presented in Table 8. Table 6. Multicollinearity Assumption Checking Variable VIF EF 1.757 DL 1.538 AU 2.059 DL*EF 1.131 Table 7. Effect Size Variable f Annotation EF 0.317 Medium DL 0.171 Medium DL*EF 0.032 Small Table 8. Path Parameter Coefficient Relationship Between Latent Variables Path Coefficients P-Value Annotation EF --> AU 0.488 <0.001 Accepted DL --> AU 0.308 <0.001 Accepted DL*EF -->AU -0.092 0.034 Accepted https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 141 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) The parameter coefficients of the SEM-PLS path in Table 8 can be explained as follows. The path parameter coefficient that marks the relationship between external factors and the ac- tual use of e-learning is 0.488 with a P-Value <0.05 (significant at the 5% level), H01 is rejected. It indicates that the external factor variable has a significant positive effect on the actual use of e- learning. learning at state Islamic universities in Indonesia during the Covid-19 pandemic. The path parameter coefficient that marks the relationship between digital literacy and the actual use of e-learning is 0.488 with a P-Value <0.05 (significant at the 5% level), H02a is reject- ed, it can be interpreted that the digital literacy variable has a significant positive effect on the ac- tual use of e-learning. learning at state Islamic universities in Indonesia during the Covid-19 pan- demic. The path parameter coefficient that marks the moderation between external factors and digital literacy on the actual use of e-learning is -0.092 with a P-Value of 0.05 (significant at the 5% level), H02b is rejected, it can be interpreted that the digital literacy variable weakens the in- fluence of external factors on the actual use of e-learning at state Islamic universities in Indonesia during the Covid-19 pandemic. The final model of this research can be seen from the path dia- gram of the SEM-PLS analysis output as presented in Figure 2. Figure 2. PLS-SEM Path Diagram (Final Research Model) The path diagram in Figure 2 shows the parameter coefficient (β) and P-Value (probability value) between the exogenous latent variable and the endogenous latent variable. The path para- meter coefficients can be entered in the mathematical model according to the structural model equation in Formula (3) as follows. The mathematical equation for the AU estimation model above can be explained as fol- lows. Parameter coefficient ( ) is 0.49, meaning; every increase in external factors by 1 point, will have a significant impact on increasing the actual use of campus e-learning by 0.49 points, as- suming other variables that influence it to remain. The parameter coefficient ( ) is 0.31, mean- ing; every 1 point increase in digital literacy will have a significant impact on an increase in the actual use of learning by 0.31 points, assuming other variables that influence it to remain the same. The parameter coefficient ( ) is -0.09, meaning; every interaction between external factors and digital literacy will have a significant impact in reducing the influence of external factors on the actual use of e-learning by 0.09 points. FINDINGS AND DISCUSSION The influence of external factors on actual use is indicated by the positive and significant coefficient of the SEM-PLS structural model parameter at the 5% level which means that exter- nal factors have a significant positive effect on the actual use of e-learning at state Islamic univer- sities in Indonesia during the Covid-19 pandemic. This finding verifies the research findings of Noh (2017) and Widana (2020) but is not in line with Davis et al. (1989) who state that external variables do not directly affect attitudes and behavior in the use of technology. https://doi.org/10.21831/reid.v7i2.44794 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 142 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) The effect of digital literacy on actual use is indicated by the parameter coefficients on the structural model which are positive and significant at the 5% level, meaning that; Digital literacy has a significant positive effect on the actual use of e-learning at state Islamic universities in Indonesia during the Covid-19 pandemic. The findings of this study are in line with the results of Jan (2017) who stated that “… digital literacy (DL), tablet and smartphone use, training, previous use of computers and frequency of computer use significantly influence students' attitudes to- wards ICT use”. The results of this study are not in line with the results of the research of Jang et al., (2020, p. 1) who found that "Digital literacy did not have a direct significant effect on the intention to use learning technology in Finland...". The Moderating effect between digital literacy and external factors is indicated by the negative and significant SEM-PLS structural path parameter coefficient at the 5% level, meaning that the digital literacy variable weakens the influence of external factors on the actual use of e- learning at state Islamic universities in Indonesia during the Covid-19 pandemic. Findings related to the moderating relationship of external factors with digital literacy show a unique phenome- non, if students are more skilled in managing information, using digital equipment, and being able to transform digital data and information, combined with the availability of adequate facilities and infrastructure, it will weaken the frequency and duration of their time in using digital literacy e- learning. This finding is interesting to observe because it contradicts the general assumption that better understanding and literacy supported by adequate facilities and infrastructure will have a positive impact on the frequency and duration of ICT use. In fact, from the results of direct ob- servations of researchers while teaching at the state Islamic universities (as participant-observers), and preliminary research interviews with PTKIN students and lecturers, that students who have more skills in the field of ICT, and adequate supporting facilities do not feel at home for a long time in using e-learning, because they can still follow the learning process through recordings of material that have been uploaded in e-learning, so most of them only record attendance online, download material to be repeated on a computer or smartphone device offline, with more time and cost-efficient reasons. The findings of this study, are supported by the results of Ferri et al. (2020) revealing that there are several challenges in using online learning media during an emergency (Covid-19 pan- demic), including technological challenges, pedagogics, and social challenges. Technological chal- lenges related to an inadequate internet connection, and lack of necessary electronic devices. Pe- dagogic challenges, related to the lack of digital skills of teachers and students, the lack of struc- tured content when compared to the number of online resources, the lack of interaction between students and teachers, and the lack of social presence and teacher cognition. Social challenges are related to the lack of interaction between teachers and students. CONCLUSION The results of this study can be concluded that; External factor variables have a significant positive effect on the actual use of e-learning. The digital literacy variable has a significant posi- tive effect on the actual use of e-learning. The digital literacy variable weakens the influence of external factors on the actual use of e-learning on Islamic college campuses in Indonesia. The researchers realize that there are several limitations in this paper, first; there is no prev- ious research that places digital literacy as a moderating variable of external factors on the actual use of e-learning. Second; The researchers only adopted and adapted the actual use variable in the Technology Acceptance Model theory developed by Davies, and added the role of digital literacy as a moderating variable from the influence of external factors on the actual use of e-learning. Third; this research was conducted in a short time, there was no instrument trial, so from the re- sults of field testing many research instrument items had to be issued. Fourth; The population limit is unknown, so the proportion of samples taken is not evenly distributed across all state Islamic universities. 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 143 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) ACKNOWLEDGMENT We would like to express our gratitude to the students of state Islamic universities throughout Indonesia who have responded to the statement of research instruments. We also ex- press our appreciation to all lecturers and education staff of the Pontianak State Islamic Institute of Religion who helped distribute research instruments to PTKIN students in Indonesia. We do not forget to express our gratitude to the lecturers and postgraduate students of Yogyakarta State University Batch of 2021 who have helped direct the preparation of this paper. REFERENCES Azzahra, N. F., & Amanta, F. (2021). Promoting digital literacy skill for students through improved school curriculum. In Policy Brief No. 11, 1-13. Center for Indonesian Policy Studies (CIPS). http://hdl.handle.net/10419/249444 Creswell, J. W. (2012). Educational research: Planning, conducting, and evaluating quantitative and qualitative research (4th ed.). Pearson. Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer technology: A comparison of two theoretical models. Management Science, 35(8), 982–1003. https://doi.org/10.1287/mnsc.35.8.982 Ferri, F., Grifoni, P., & Guzzo, T. (2020). Online learning and emergency remote teaching: Opportunities and challenges in emergency situations. Societies, 10(4), 86. https://doi.org/10.3390/soc10040086 Garson, G. D. (2016). Partial least squares: Regression & structural equation models. Statistical Publishing Associates. Hair, J. J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2014). A primer on partial least squares Structural Equation Modeling. Long Range Planning, 46(1), 184-185. https://doi.org/10.1016/j.lrp.2013.01.002 Hermawan, D. (2021). The rise of e-learning in covid-19 pandemic in private university: Challenges and opportunities. IJORER: International Journal of Recent Educational Research, 2(1), 86–95. https://doi.org/10.46245/ijorer.v2i1.77 Jan, S. (2017). Investigating the relationship between students’ digital literacy and their attitude towards using ICT. International Journal of Educational Technology, 5(2), 26-34. https://ecommons.aku.edu/pakistan_ied_pdck/304/ Jang, M., Aavakare, M., Kim, S., & Nikou, S. (2020). The effects of digital literacy and information literacy on the intention to use digital technologies for learning - A comparative study in Korea and Finland. In ITS Online Event, 1-13. International Telecommunications Society. http://hdl.handle.net/10419/224858 Kaeophanuek, S., Na-Songkhla, J., & Nilsook, P. (2019). A learning process model to enhance digital literacy using Critical Inquiry through Digital Storytelling (CIDST). International Journal of Emerging Technologies in Learning (IJET), 14(03), 22–37. https://doi.org/10.3991/ijet.v14i03.8326 Kock, N. (2019). WarpPLS user manual: Version 6.0. ScriptWarp Systems. https://scriptwarp.com/warppls/ Lemeshow, S., Jr., D. W. H., Klar, J., & Lwanga, S. K. (1990). Adequacy of sample size in health studies. WHO. https://doi.org/10.1002/sim.4780091115 10.21831/reid.v7i2.44794 Sumin, Kahirol Mohd Salleh, & Nurdin Page 144 - Copyright © 2021, REiD (Research and Evaluation in Education), 7(2), 2021 ISSN: 2460-6995 (Online) Liu, Z.-J., Tretyakova, N., Fedorov, V., & Kharakhordina, M. (2020). Digital literacy and digital didactics as the basis for new learning models development. International Journal of Emerging Technologies in Learning (IJET), 15(14), 4–18. https://doi.org/10.3991/ijet.v15i14.14669 Lohmöller, J.-B. (1989). Latent variable path modeling with partial least squares. Physica-Verlag HD. https://doi.org/10.1007/978-3-642-52512-4 Noh, Y. (2017). A study on the effect of digital literacy on information use behavior. Journal of Librarianship and Information Science, 49(1), 26–56. https://doi.org/10.1177/0961000615624527 Putro, S. T., Widyastuti, M., & Hastuti, H. (2020). Problematika pembelajaran di era pandemi COVID-19 studi kasus: Indonesia, Filipina, Nigeria, Ethiopia, Finlandia, dan Jerman. Geomedia: Majalah Ilmiah Dan Informasi Kegeografian, 18(2), 50–64. https://journal.uny.ac.id/index.php/geomedia/article/view/36058 Rosenberg, B., & Navarro, M. A. (2018). Semantic differential scaling. In B. B. Frey (Ed.), SAGE encyclopedia of research, measurement, and evaluation. SAGE Publications. https://scholar.dominican.edu/books/144 UNESCO. (n.d.). Education: From disruption to recovery. Retrieved August 3, 2021, from https://en.unesco.org/covid19/educationresponse Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 46(2), 186–204. https://doi.org/10.1287/mnsc.46.2.186.11926 Vinzi, V. E., Chin, W. W., Henseler, J., & Wang, H. (2010). Handbook of partial least squares: Concepts, methods and applications. Springer Berlin Heidelberg. https://doi.org/10.1007/978-3- 540-32827-8 Widana, I. W. (2020). The effect of digital literacy on the ability of teachers to develop HOTS- based assessment. Journal of Physics: Conference Series, 1503(1), 012045. https://doi.org/10.1088/1742-6596/1503/1/012045