The notion of self-efficacy is defined as the judgements that a subject makes on their own abilities to organize and execute courses of action necessary to achieve different objectives ( Bandura, 1986, 1989 ). Many factors exist that determine human behavior; Bandura (1997) identifies self-efficacy as an important mechanism that influences both individuals’ performance in task execution and cognitive processing. In accordance with the foregoing, it is possible to assert that the mental construction that individuals make of their own self-efficacy has a variety of distinctive characteristics. These are important because they allow distinguishing it from other psychological notions; additionally, these include implication on how self-efficacy should be measured.

In the first place, self-efficacy has specific attributes through activities and contexts, which are: level, generality, and strength. Level refers to the difficulty of one task in particular, generality refers to the transference of the beliefs of self-efficacy of one task to another, and strength is determined by the amount of certainty that the individual has to successfully complete a determined task ( Bandura, 1989 ). In the second place, self-efficacy beliefs are not a unique disposition, they are multidimensional in form and differ on the level of the task or domain ( Schunk, 1991 ). In the third place, regarding content, self-efficacy measures are focused on subjects’ performance or abilities more than on personal qualities, such as physical development or psychological characteristics ( Bandura, 1986 ).

Bandura posits the existence of four main sources of information that determine individuals’ self-efficacy: domain experiences, vicarious experiences, verbal persuasion and social influence, and physiological states ( Bandura, 1986: p. 424 ). Of the four sources, domain experiences are the most effective way of creating a strong sense of self-efficacy ( Bandura, 1997 ). These refer to the interpretations that individuals make of their past actions and they contribute the most authentic proof of the abilities possessed to successfully achieve the development of a learning task ( Usher & Pajares, 2009 ). The second source of self-efficacy are vicarious experiences or indirect experiences. “Viewing or imagining that other similar individuals act successfully, possibly increases the observer’s self-perception of self-efficacy, coming to believe that they themselves also possess the abilities” ( Bandura, 1986: p. 425 ). Indirect experiences are not as strong as domain experiences, however, these can produce significant and lasting changes in self-efficacy through their effect on performance ( Bandura, 1986 ).

Verbal persuasion and social influence is the third source of self-efficacy. It is used to induce individuals into believing that they are capable of reaching the objective laid out. Individuals that are convinced verbally by others that they possess the ability to carry out the tasks are susceptible of mobilizing a greater effort and maintaining it when faced with difficulties ( Schunk, 1991 ). The last source of self-efficacy are the physiological states. Levels of moderate physiological activation (sweating, agitation, fatigue, etc.) will facilitate the deployment of abilities, while an elevated activation will influence the levels of self-efficacy.

In recent decades, an important number of studies have developed around self-efficacy in the field of ICT and education. In fact, diverse research has examined this notion from different perspectives according to Bandura (1997) , who asserts that self-efficacy beliefs can be specified in the level of the task or domain. In accordance with this, and before presenting the results of the bibliometric study, it is convenient to identify the different fields of study of self-efficacy in association with ICT as a domain, which are found in the research that are analyzed in the present study.

The first field of study, academic self-efficacy, has been the object of study during the execution of learning tasks in computational scenarios ( Jan, 2015 ). It refers to students’ perception of their ability to complete academic tasks in these scenarios ( Girasoli & Hannafin, 2008 ). Research reveal that students with high academic self-efficacy participate more actively in educational activities, make more of an effort, and can develop effective strategies when obstacles arise ( Odaci, 2013 ). It is noteworthy to mention that the factors that influence individuals’ AS derive from both the computational learning environment and the topic of study ( Moos & Azevedo, 2008 ).

The second field of study, computer self-efficacy, arises during the 90s with the boom of computers in educational contexts. It is defined as individuals’ general beliefs about their abilites to use computers competently through multiple domains ( Hasan, 2006 ). For Torkzadeh, Chang, & Demirhan (2006) , computer self-efficacy plays an important role in an individual’s decision to use a computer and its resources and, therefore, in the disposition to learn other knowledge related to the use of this device.

Computer self-efficacy and students’ performance has also been object of study. For example, Compeau & Higgins (1995) assert that computer self-efficacy exercises significant effects on result expectations and performance in computational scenarios. Peinado & Ramírez (2014) found that students with a high level of computer self-efficacy exhibit better attitudes and academic achievements than their peers with a low or medium level of computer self-efficacy. According to the foregoing and to the findings of Moos & Azevedo (2009) , behavioral and psychological factors have a positive relationship with computer self-efficacy, which, in turn, positively influences students’ learning results in computational environments.

The third field of study, referred to as the field of interest, is self-efficacy in internet-based learning environments. Is defined as the judgements that students make about their ability to organize and execute activities related to the Internet in order to produce the desired results ( Eastin & LaRose, 2000 ). More recently, Cheng & Tsai (2011) redefine it as students’ expectations and confidence to participate and learn in Web-based environments. Diverse studies assert that the increase in levels of self-effi- cacy in internet-based learning environments can lead to the development of better cognitive and metacognitive abilites for information searches in a Web-based environment and, in turn, promote students’ learning ( Tzeng, 2009 ).

Research into self-efficacy in internet-based learning environments has been more prominent than in the foregoing postulates. In this regard, Tsai et al. (2011) asserts that IS has been linked to: learning processes ( Lu et al., 2007 ), learning results ( Chu & Tsai, 2009 ), anxiety ( Ekizoglu & Ozcinar, 2010 ), results expectations ( Bates & Khasawneh, 2007 ), and information searches ( Chiou & Wan, 2007 ), among others.

Based on the foregoing, it is possible to assert that in the research into the role that individuals’ self-efficacy plays when they interact with computational learning environments, this has been researched significantly in recent decades ( Tsai et al., 2011 ). A practical evolution of the concept can be observed, which has allowed researchers to effectively link it as a variable object of study, which, in turn, derived into the establishment of didactic strategies for the design of computational scenarios that take into account students’ motivational dimension ( López, Hederich, & Camargo, 2012 ). Consequently, the present study has gathered and studied 81 research articles published between 2006 and 2015, derived from the fields of study already described, and it performs a bibliometric analysis of the literature in question.

Regarding the origin of the scientific papers, it was found that they proceed from 19 countries. In terms of production by country (Figure 1), the countries that stand out for exhibiting greater research and productivity in the subject matter are Taiwan, with 23 papers; United States, with 17, and Turkey, with 11. These papers, as a whole, represent 63% of total papers. Asian countries contribute 49% of the research in the area of interest, followed by North America, with 21%; Europe, with 20%, and, finally, South America, with 10%.

In relation to the contexts within which the studies are conducted and participants’ levels of education, higher education stands out as the context within which the subject matter is studied the most. In Figure 2, a representation of the situation is shown. Indeed, of the 81 articles found, 48 (62%) refer to applications in universities. With much smaller shares, there are applications in secondary and middle education 12 (15%), basic primary education 8 (10%), distance education through Internet 7 (9%), and adult education 3 (4%) in informal contexts. It is noteworthy to mention that these values do not take into account 3 studies of a documentary nature.

Figure 3 shows the distribution of participants only in the context of higher education by discipline. As it can be observed, participants that come from higher education programs that educate educators represent 38% of the population in this context and, in turn, these represent 22% of the total of studies on the subject matter. There are also participants from nursing and medicine programs (11%), engineering (8%), and psychology (8%). It is noteworthy to mention that 18% of the participants come from diverse disciplines, such as social sciences, humanities, and architecture among others.

On the other hand, Figure 4 shows the types of computational environments that researchers use to develop the studies, which are the scenarios with which participants interact. E-learning environments are the most used: courses embedded in Learning Management System (LMS) platforms, mainly Moodle and Blackboard. These are followed by the Internet and to a lesser degree, hypermedia environments.

Figure 5 shows the domains of knowledge that frame the development of research activities, according to the evidence gathered. As it can be observed, the field where most publications converge is Computer Science, with 63% (digital literacy, Internet,

hardware, and software), followed by Science and Technology (15%), and to a lesser extent Mathematics (8%), Languages (5%) and Education and Psychology (4%).

Table 1 shows the number of publications found in this review by year of publication. The year 2010 is the period with the most publications on the subject matter with a total of 13 articles, followed by the years 2011, 2013, and 2015, each one with 10 articles. In contrast, the years 2007 and 2009 have the lowest number of publications: 5 articles per year. An important increase can be observed in the number of publications starting from the year 2010 and up to the year 2015, this period represents 70% of total papers. An increase can be observed in the number of publications in the last five years as a result of the heightened academic interest in the subject matter of self-efficacy associated with the use of a computer.

Articles on the subject matter located in this study have been published in a total of 31 refereed scientific journals. Table 2 shows the list of the journals’ names, the amount

of articles published on the subject matter during the period of observation, indexing type according to the 2015 Colombian National System of Indexing and Official Approval of Specialized Journals of Science, Technology, and Innovation (PUBLINDEX) and the impact factor. Among the journals that stand out are two from the United Kingdom, Computers & Education, with 27 publications on the subject matter, and Computers in Human Behavior, with 11 publications. These journals are the basis of the high productivity in the field of the present study. The rest of the journals have between 1 and 3 publications, with the exception of Procedia-Social and Behavioral Sci- ences, which has 7 publications.

Table 3 shows the relationship between the number of articles and the number of citations per year. A total of 3317 citations were found during the period of observation, of which 2744 citations correspond to the set of publications yielded by the Science Direct database, 532 correspond to Scopus, and 9 to SciELO. The year 2006 is the period with the largest number of citations, with 744 citations in 7 articles; followed by the year 2010, with 587 citations, and the year 2008, with 554 citations. The year 2015 had the lowest number of citations.

Figure 6 shows the indexes of references per year. This bibliometric indicator is obtained based on the quotient between the number of references and the number of published articles per year. It is an indicator of the use of scientific literature for a specific period of time. As can be observed, the index of references peaked in the year 2006, with a value of 106.3, followed by the year 2007, with 90.6, and the year 2008, with 79.1. These values are consistent with the time that the articles have been in circulation and available to the academic community.

The number of citations received by each article was obtained through Google Scholar. The range is between 0 and 175 citations. The most cited articles are “The relationship of e-Learnerʼs self-regulatory efficacy and perception of e-Learning environmental quality” by Lee and Lee (2008) , with 175 citations; “Computer use and the gender gap: The issue of access, use, motivation, and performance” by Imhof, Vollmeyer, & Beierlein (2007), with 169 citations; and “University Students’ Internet Attitudes and

Internet Self-Efficacy: A Study at Three Universities in Taiwan” by Wu & Tsai (2006) , with 159 citations, during the 2006-2015 period.

In the 81 studies analyzed in this review, 194 authors were identified. Table 4 shows the relationship between widely published authors in the subject matter and the number of articles in which they have participated, whether in an author or coauthor condition. Sixteen authors have participated in two publications; two authors participated in three publications; one author participated in four publications; and one author participated in eleven publications. As can be observed in Table 4, Dr. Chin-Chung Tsai, professor at the Graduate Institute of Digital Learning and Education of the National Taiwan University of Science and Technology, is the most cited author, with 11 publications on the subject matter.

Similarly, Table 5 shows the coauthorship index or signatories/article per period. This Bibliometric indicator refers to the quotient between the total number of signatories (authors) and the number of published articles per year. As can be observed, the authors’ mean oscillates between 2.0 authors, corresponding to the years 2009 and 2012, and 2.8 authors, in the years 2007 and 2013.

With respect to research methods, Table 6 shows those identified. Correlational type studies are predominant, with a frequency of 51 studies (63%), followed by studies of an experimental nature, with a frequency of 14 studies (17%), and quasi-experimental nature, with 7 studies (9%); all of these with a quantitive approach. Descriptive and documentary type studies are used to a lesser extent, each one with three studies, and the exploratory type studies and case study, each one with one study. These results are similar to those reported by Tsai et al., (2011) , who in a literature review study during the period between 1999 and 2009 found, in the set of studies analyzed (46 in total), that research with a quantitive and correlational approach prevails.

In regard to the instruments used in the studies, a significant and important number of scales and questionnaires were found, which are used and elaborated by the authors themselves; in some cases, as information gathering sources. A total of 62 questionnaires to measure self-efficacy were counted, which are characterized for being self- reporting, Likert-type scale, whose purpose is to evaluate participants’ perception with respect to their self-efficacy in different domains. Among the latter, those mainly associated with self-efficacy with the use of a computer, with the use of Internet and Web or online learning environments, and academic self-efficacy, were identified.

Table 7 shows the most frequently used instruments in the studies considered. Those that stand out because of their use in research are: the Self-efficacy subscale of the MSLQ ( Pintrich, Smith, Garcia, & McKeachie, 1991 ), which is used in 14 studies, followed by the Internet Self-efficacy Scale (ISES) and the Web-based Learning Self-Effi- cacy Scale (WBLSES) ( Tsai & Tsai, 2003 ), used in 9 studies, and the Computer Self-ef- ficacy Scale (CSES) ( Compeau & Higgins, 1995 ), used in 4 studies.