Massive open online courses in higher education : A data analysis of the MOOC supply

Purpose: The aim of this study is to analyze the factors influencing the MOOC supply level. Specifically, this paper analyzes certain internal and strategic factors associated with universities, such as prestige, public or private status, age, size (measured by the number of faculty members or students) and region. Design/methodology: We apply a descriptive methodology and then use multivariate analysis to test five hypotheses related to the institutional profile of 151 universities in 29 countries. Empirical evidence is provided from universities offering MOOCs through the four of the most commonly used private global platforms that emerged as part of the booming MOOC movement (Udacity, Coursera, edX and MiríadaX). Findings: The findings show some differences when prestige is measured according to the Shanghai ranking (model 1) and the Webometrics ranking (model 2). In both cases, the private nature of the university and the region (North America) are factors that have a significant influence on the MOOC university supply. Depending on both rankings, size and age are influential factors. It is important to emphasize that prestige is a significant factor according to model 2.


Introduction
The University, as a social institution called upon to respond to the challenges of modernization and globalization, has incorporated many forms of information and communications technologies (ICT).
As a result, new virtual educational settings have been generated, which have meant important transformations, ranging from the way knowledge is accessed to the very structure of the educational institutions (European Commission, 2013;UNESCO/COL, 2011).
The effects and trends of ICTs in higher education have been analyzed for more than a decade (Kirkup & Kirkwood, 2005;Alba Pastor, 2005).Taylor and Osorio (2005) predicted for this decade that North American university education would be firmly established as an export product to the world, based on the use of ICTs.This prediction seems to have become true with Massive Open Online Courses (MOOC).In fact, it is believed that the technological advances in higher education were not overly visible until the emergence of these courses (EUA, 2015), the rise of which since 2012 has been the subject of great debate in today's educational and pedagogical field (Ng'ambi & Bozalek, 2015;Sangrà, González & Anderson, 2015).Their importance lies in the proposal of a barrier-free education for anyone who has an Internet connection, thanks to their free and open characteristics, which are aspects that promote another of their main features: their massive access.
While they are a phenomenon originating in North American universities (Rodriguez, 2012;Waldrop, 2013), MOOCs do not form part of the formal university course offering and are taught via virtual education platforms that have, for the most part, been created as consortia for this purpose (e.g., Coursera, Udacity, edX, MiríadaX), which are offered by different non-and for-profit entities (Ong & Grigoryan, 2015;Pang, Tong & Na, 2014;Yuan & Powell, 2013).Some studies (Daniel, Vásquez & Gisbert, 2015;Yuan & Powell, 2013) consider the participation of the universities in the MOOC phenomenon to be the result of the technology-media convergence process and the consequence of the massification of higher education in the context of the cultural homogenization that is associated with globalization.Others consider MOOCs to be an opportunity for public universities with smaller budgets, and not the threat that was originally imagined, pointing out their benefits in terms of reaching social groups such as retirees or employees who wish to improve their technical performance, for whom these courses would be an intellectual challenge (Ong & Grigoryan, 2015).They are also seen as an opportunity to advance in lifelong learning (De Freitas, Morgan & Gibson, 2015), a view that is shared by official European bodies, which consider them to be a change agent in higher education (European Parliament, 2015;European Commission, 2013).Two characteristics have set the MOOCs apart from the already well-established e-learning industry: their massiveness and openness (Atenas, 2015), mostly in the sense of their free cost rather than in the original sense of "open educational resources" (OER).However, these basic principles of openness, reuse and recombination (OECD, 2015) have been obscured by the fact that MOOCs might be initially free-of-charge, but the suppliers could sometimes add charges for additional services, such as accreditation and certification (Daniel et al., 2015;Atenas, 2015).This has resulted in the criticism that associates the boom of the MOOCs with economic and commercial gain.
Most of the research on MOOCs has focused more on aspects related to their demand, and less on their offering (Ospina & Zorio, 2016;Gašević, Kovanović, Joksimović & Siemens, 2014).In general, it is taken for granted that the educational institutions offer this type of courses mainly to expand their scope and institutional visibility, to attract more students and to build and maintain their brand (Sangrà et al., 2015;Daniel et al., 2015).These reasons have been identified in North American studies, and works have also recently been published based on questionnaires in Europe, which have added to the references on the offering and the development perspectives of MOOCs.According to the data from the European University Association, MOOCs represent one of the activities with the greatest growth potential in European universities (EUA, 2015).The Conference of Presidents of Spanish Universities (CRUE, 2015), in turn, indicates that the reasons why universities offer these courses are, first of all, innovation in learning, and secondly, visibility and the presence of university teaching on the Internet.
The novelty and timeliness of our study is highlighted by the recent research presented above, as we aim to analyze the MOOC supply through the four most popular platforms.Our findings identify the characteristics of the universities that are actively taking part in this educational trend.To begin with, five hypotheses are tested that are associated with the prestige, type (public or private), age, size and region of the university, and then a multivariate analysis is carried out to identify whether any of these characteristics of the universities has an influence on the MOOC offering.This work thus contributes new evidence of the empirical relationships that exist among variables associated with the profile of the universities and their level of MOOC offering, and in particular, shows the relevance of prestige as a strategic factor.Following this introduction, the next section presents a review of the international literature on MOOCs and the formulation of the study hypotheses.The third section describes the data obtained, as well as the variables and the methodology used.Next, in the fourth section, an analysis of the results is presented, followed by a final section with the main conclusions derived from the work.

Theoretical framework and hypotheses development
The acronym MOOC is used generically to refer to two types of courses with different educational foundations (Sangrà et al., 2015;Clow, 2013).C-MOOCs are based on the educational principles of the "connectivism" proposed by Siemens (2004) as a theory of learning for the digital age, emphasizing the power of social interactions to autogenerate knowledge.Meanwhile, x-MOOCs are based on contents and the traditional concept of knowledge transmission and are more commonly found on the earliest and largest educational platforms that supply MOOCs such as Coursera, Udacity, edX and MiríadaX (Atenas, 2015;Rodriguez, 2013), on which this present work is focused.
From the perspective of the theory of disruptive innovations by Bower and Christensen (1995), MOOCs are described as an innovative practice involving an educational disruption (Yuan & Powell, 2013;Anderson & McGreal, 2012), and more precisely, a disruptive innovation of the market, by becoming an interesting product that combines a technological development with a new business market that promotes a more flexible type of low-cost training (Vásquez, López & Sarasola, 2013).
In spite of the novelty of the topic, the body of literature is continuously expanding (Sangrà et al., 2015;Gašević et al., 2014;Liyanagunawardena, Adams & Williams, 2013a), especially in terms of research on the MOOC demand-side.Christensen, Steinmetz, Alcorn, Bennett, Woods and Emanuel (2013) report that most of the participants in these courses are young people with higher education and employees from developed countries, who are both looking to satisfy their curiosity and improve their career profile.Yousef, Chatti, Wosnitza and Schroeder (2015) have grouped the reasons why students participate in MOOCs into eight clusters: mixed learning, flexibility, high-quality content, instructional design and learning methodologies, life-long learning, on-line learning, openness and student-focused learning.
Taking the perspective of the supply side, Hollands and Tirthali (2014)

look into why institutions offer
MOOCs, with a qualitative study of 83 interviews with leaders of 29 US institutions.They identify six main objectives: expanding the institutional scope and attracting a larger number of students (size), building and maintaining their brand (prestige), improving their finances by reducing costs or increasing income, improving their educational results, innovating in teaching-learning and conducting research on teaching and learning processes.Meanwhile, the study by Jansen, Schuwer, Teixeira and Aydin (2015), based on online surveys of 67 institutions of higher education in 22 European countries that offer MOOCs or that plan to do so indicates that, unlike in the United States, the most important objective is to create new opportunities for flexible learning.It also finds that the number of European universities offering MOOCs or that plan to do so increased from 58% in 2013 to 71.7% in 2014, while in the USA, according to the data from Allen and Seaman (2015), it decreased from 14.3% to 13.6%.Thus, Jansen et al. (2015) have concluded that the European universities seem to be more committed to the MOOC phenomenon than their US counterparts, to the point of becoming a major trend in Europe.
In Spain, a study sponsored by Telefónica, (Oliver,Hernández,Daza,Martin & Albó,2 014) with data prior to December 2013 indicates that 35% of the universities have at least one MOOC and that the phenomenon involves both universities with a tradition of distance learning as well as traditionally onsite universities.It also identified public universities as having a more extensive offering and Spain as the leader in MOOC offerings in Europe, a fact that has also been acknowledged by the CRUE (2015).
In this framework, this paper focuses on studying whether certain characteristic factors of an institution (prestige, type of university [public or private], age, size and region of origin) explain the level of university offering of MOOCs.

Prestige
Within the current dynamics of the internationalization of higher education, one of the most important common objectives among universities is to increase their reputation and prestige; as part of this, their position in the rankings plays a key role (European Parliament, 2015).As a matter of fact, in a competitive environment, reputation and institutional prestige are one of the factors that determine, among other things, the selection of a university and degree program by students (Sierra, 2012;Aguillo, Bar-Ilan, Levene & Ortega, 2010;Docampo, 2008).It has even been confirmed that prestige does not necessarily derive exclusively from strictly research activities (Horstschräer, 2012;Maringe, 2006).
For this reason, prestige has been analyzed as a strategic factor of universities (Hollands & Tirthali, 2014;Jordan, 2014;Ospina & Zorio, 2016), and is a term frequently used by MOOC platforms, which declare that their offering comes from the world's most prestigious universities (Ong & Grigoryan, 2015;Schuwer et al., 2015;Pang et al., 2014;Yuan & Powell, 2013).Reinforcing their own prestige is also one of the objectives sought by MOOC participants, according to Yousef et al. (2015).Therefore, it could be considered that the universities with the greatest prestige lead the offering with a larger number of MOOCs, as a clear sign of their commitment to educational innovation.Therefore, the first hypothesis proposed is: H1: There is a significant positive relationship between the prestige of a university and its MOOC offering.

Public or private status
While it is true that MOOCs are a phenomenon in which both public and private universities participate, the existing literature associates it primarily with the interest of private agents, with substantial levels of investment (Yuan & Powell, 2013;Belleflamme & Jacqmin, 2014), due to a large extent to the growth trend in private education (UNESCO, 2009).With regard to US institutions, Hollands and Tirthali (2014) and Allen and Seaman (2014) indicate statistical differences between the types of institutions offering MOOCs, in which private institutions predominate.Likewise, Shrivastava and Guiney (2014) emphasize in their study a greater trend towards these courses in private universities.This is not the case in Spain, where there is a greater offering by public universities, which have benefited from the impulse of one of the most commonly used platforms, "MiríadaX", which is the product of private alliances (Oliver et al., 2014).To validate this trend, the following hypothesis is proposed: H2: Private universities have a greater offer of MOOCs than public universities.

Age
The age of the university has been analyzed in different studies on higher education strategies (Luque-Martínez, 2013;Guzmán, del Moral, González & Gil, 2013;Gallego-Álvarez, Rodríguez & García-Sánchez, 2011), which have found, for example, a significant negative relationship between the age of the university and the use of ICT (Iniesta, Sánchez & Schlesinger, 2013).The oldest universities, which have built their know-how over many years, tend to safeguard their corporate image more than younger institutions (Gallego-Álvarez et al., 2011) and as a result, they may be less enthusiastic about models whose effectiveness is still in the early stages, such as MOOCs.Based on this, the following hypothesis has been developed: H3: There is a negative relationship between the age of a university and its MOOC offering.

Size
The size of the institution, measured in terms of the number of students or faculty members, is another factor analyzed in higher education literature (UNESCO, 2009;Luque-Martínez, 2013;Guzmán et al., 2013;Gallego-Álvarez et al., 2011).According to the study by Allen and Seaman (2014), larger US institutions, with more than 15,000 students, are more likely to offer MOOCs.Other studies (Hollands & Tirthali, 2014;SCOPEO, 2013) indicate that, in many cases, the initiative is a response to alleviate the pressure of the high-demand for "bottleneck" courses, a factor associated with the size in terms of the number of students.According to Ospina and Zorio (2016), a small size, measured in terms of the number of faculty members, is one the only condition present in one of the two solutions leading to non MOOC-intensiveness.Hence, we put forward that larger universities are the ones that have a greater level of resources to deal with this type of innovation, and thus the following hypothesis is proposed: H4: There is a positive relationship between the size of a university and its MOOC offering.

Region
A study by Aguillo, Ortega and Fernández (2008) has led us to consider the importance of this factor, by suggesting a trend on the part of US universities to make better, more in-depth use of the Internet.
Furthermore, even though it has been acknowledged that the first MOOC was launched in Canada in 2008, and the literature indicates the preponderance of the phenomenon in the USA (Schuwer et al., 2015;Yousef et al., 2015;Shrivastava & Guiney, 2014;Liyanagunawardena et al., 2013a;SCOPEO, 2013), new participants in other regions are joining the offering, indicating even a greater, growing trend related to this phenomenon in Europe (Jansen et al., 2015).As a result, it would be interesting to validate this aspect through the following hypothesis: H5: US universities have a greater offer of MOOCs than the universities from the rest of the world.

Study design: Methodology, sample and variable definitions
The methodology consists, firstly, of a descriptive statistical analysis for an initial approximation to the study variables, followed by a bivariate analysis that makes it possible to explore the importance of each of the variables available in relation to the MOOC offering.Secondly, a multiple regression analysis enables us to corroborate which of the variables included in the model have some degree of effect on the level of MOOC offering.

Sample
There are several universities that offer MOOCs through either their own virtual e-learning platforms or pre-existing Learning Management Systems (LMS), such as Blackboard or Moodle (Ong & Gregorian, 2015), but for the purposes of this study, we are interested in those that have decided to offer them through one of four external platforms.Our sample includes 151 universities from 29 countries, organized into four large regions, and with a total offering of 904 courses (see Table 1).The universities correspond to those that on June 30, 2014 offered at least one MOOC on one of the first global platforms created by private initiative, and which have had the greatest visibility and distribution during the boom of the MOOC movement: Udacity, Coursera, edX and MiríadaX (Sangrà et al., 2015;Jordan, 2014;Clow, 2013;Anderson & McGreal, 2012;Daniel, 2012).MOOCs offered by providers other than universities were excluded (Raposo, Martínez & Sarmiento, 2015).

Variables
To measure prestige (H1), we use two rankings that are widely recognized around the world, but that have substantial methodological differences between them: Shanghai (ARWU, 2014), whose classification prioritizes research activity (Aguillo et al., 2010;Aguillo et al., 2008;Docampo, 2008) and Webometrics (Cybermetrics Lab, 2014), which classifies the universities according to their visibility on the Internet (Chen, Tang, Wang & Hsiang, 2015;Garde-Sánchez, Rodríguez & López, 2013;Aguillo et al., 2008).Using a qualitative approach (fsQCA), Ospina and Zorio (2016) demonstrate that the absence of prestige, measured through the Webometrics ranking, is a sufficient condition to lead to a non-intensive MOOC profile by universities.The Shanghai ranking is also included in the present study to check whether, like Webometrics, it is related to the MOOC offering.
The SHANGHAI variable has thus been assigned the value of 0 when the university is not present in the ranking, and the value of 1 if it is included.The WEBOMETRICS variable has been assigned its value according to four intervals, depending on each university's classification in the ranking, with 1 meaning that the university holds one of the last positions (above one thousand) and 4 when it holds one of the top positions (in the first one hundred positions).For H2, the variable TYPE has been given a value of 0 when the university is public and a value of 1 when it is private.The hypothesis concerning the age of the university (H3) is measured by the variable LNAGE, which constitutes the Napierian logarithm of the age of the institution in years since it was founded.To measure the size of the university (H4), in line with previous studies (Allen & Seaman, 2014;Guzmán et al., 2013;Gallego-Álvarez et al., 2011), two variables have been used, LNFACULTY and LNSTUD, which represent the natural logarithm of the number of faculty members associated with the institution and the number of students registered at the university, respectively; the bivariate analysis makes it possible to select the variable between the two that offers closest relationship and the highest level of significance with the dependent variable.The variable REGION (H5) groups the country of origin of each university (Guzmán et al., 2013) into four large regions: North America, Central and South America, Europe and Asia, and Oceania.
The GDP variable is assigned the value of 0 when the GDP per capita of the university's country is less than the mean of the 29 countries in the sample, which in this case is USD $50.000, and the value of 1 when if greater.The variable INTERNET takes the value of 0 when the Internet penetration (% of the population with access) in the country of the university is less than the mean of the countries in the sample, which in this case is 70%, and the value of 1 if larger.
To analyze the relationship of the MOOC offering according to the described variables, two models of multiple regression are proposed: • incorporating the SHANGHAI variable as a measure of prestige more focused on the impact of scientific production, and on the other hand, • incorporating the WEBOMETRICS variable, which is focused mainly on the overall impact of the university on the Web.
In both cases, only one of the variables associated with size is considered: where MOOC is the dependent variable that represents the number of MOOCs offered by the University; "i" represents each university included in the sample; β0 is the constant term parameter; and each of the coefficients of the variables is represented by β1,… β7.The error term, ε, accounts for the factors not observed in the model.The program Stata v.12 was used for the statistical processing of the data.

Analysis of the results
Below is an analysis of the descriptive statistics, followed by the bivariate analysis, and finally the multiple regression analysis.

Descriptive analysis
Tables 2 and 3 present the descriptive statistics for all of the variables obtained.The first relevant aspect shown by Table 2 is the high level of dispersion of the dependent variable MOOC (S.D.=6.3), which is due to the presence of many universities offering few courses.The minimum value 1 corresponds to 30 universities (20%) that as of the study date only had one MOOC (the lowest offer), while very few universities had a high offer of Moocs, with only one university reaching the maximum value of 32 courses (the greatest offering; 0.7%).Among those with the least offering, 70% are public, 47% are European and 20% are North American; the university with the maximum offering is private and North American.In the 25th percentile, with two MOOCs, are 47 universities (30%) and in the The mean age of the universities is 156 years; 81% of the total are less than 200 years old and only 10% have an age of between 300 and 600 years, with two European (specifically, Spanish) universities having the minimum and maximum values.Of the North American institutions (60), 57% are below the mean, with the youngest being 49 and the oldest 378 years old.
With regard to size, the minimum value for both faculty (108) and students (166) is held by the same private North American university, while the maximum value for faculty (37,610) corresponds to a public Central American university and the maximum number of students (459,550) is held by a public North American university.In terms of prestige, 66.2% of the universities in our sample are in the Shanghai ranking, and they provide most of the MOOC offering (77.4%); likewise, 71.5% of the universities fall within the first 500 positions on the Webometrics ranking (categories 3 and 4 of the variable), accounting for 83% of the courses.
Most of the universities (69.5%) are public and provide 62.6% of the offering, however, of them, 71% have an offering that is below the mean, i.e., they offer fewer than six courses.
With regard to geographic region, 39.8% of the universities are in North America and account for 56.6% of the offering; 36.4% are located in Europe and offer 29% of the courses; 15.9% belong to the regions of Asia and Oceania, with 11% of the MOOCs; and the Central and South American region provides 7.9% of the universities and 3% of the offer.
Furthermore, 57% of the universities analyzed, which participate with an offering of 68% of the MOOCs, belong to countries whose GDP per capita is greater than USD $50,000, while 85% of the universities, representing 90% of the MOOC offering, are form countries with an Internet penetration greater than 70%.

Bivariate analysis
Table 4 shows the correlations between the continuous independent variables and the MOOC offering variable.Even though the correlation coefficients of the two size variables (LNFACULTY and LNSTUD) with the dependent variable are not high, they show a 5% level of significance.This first analysis suggests that the age of the university is not related to its MOOC offering, however, considering the approach of our hypothesis, H3, it must not be ruled out for the following analysis.The contrast tests for the categorical variables, shown in Table 5, reveal a high level of statistical significance (1% and 5%) for all variables, except TYPE.This includes the control variables, which suggests the importance of these variables in the analysis of the level of university MOOC offering.

Multivariate analysis
In order to avoid autocorrelation among the variables referring to size, the variable LNSTUD, which presented the lowest correlation with the dependent variable, was excluded from the regression analysis.
The absence of multicolinearity problems is confirmed by the variance inflation factor, which was lower than 10 (VIF=2.44 and 2.41) in the two models.For the variable REGION, the results are shown according to the top level in the category, i.e., Asia and Oceania.Both models (Table 6) reached overall significance (heteroscedasticity-robust F statistic=0.000) and a moderate level of predictive power, which was lower in the model (1) estimated with the SHANGHAI variable (R2=0.192)than in that estimated with WEBOMETRICS (R2=0.267).Although it is not evident in the table, the inclusion of the control variables resulted in better results in terms of significance in both models.

Independent Variables
Model The results of the regression for the model (1), with the SHANGHAI variable, indicate that the variables which influence the MOOC offering are TYPE (p<0.10), for which the coefficient indicates a three course advantage for private universities over public universities; LNFACULTY (p<0.05), which indicates that as this variable increases, so does the unit representing the level of offering; and REGION, specifically, North American with a p-value<0.01.Model (2), which uses the variable WEBOMETRICS to measure prestige, indicates that the variables which influence the MOOC offering are: WEBOMETRICS (p<0.01), as a university's position in the ranking moves up one level according to our categories, its offering increases by 2 courses; the variable TYPE (p<0.05), which reflects a higher level of significance and a larger coefficient than in model ( 1), but also indicates the influence of private university status as compared to public status in terms of MOOC offering; LNAGE, which shows a significant negative relationship (p<0.10) between the age of the universities and their MOOC offering; and the variable REGION, in which, in addition to North America (p<0.01)as in model ( 1), Europe also presents a significant p-value<0.05,which demonstrates that North American universities have a 4 course advantage over the Asia and Oceania region, while in the case of European universities, the positive difference is 2 courses as compared to the reference region.
In both models, the common variables that reveal a significant positive influence on MOOC offering are TYPE and REGION (North America), which would suggest that H2 and H5 should be accepted.
The variable LNFACULTY reveals an influence only in model ( 1), and thus H4 could be partially accepted.Likewise, judging from the results of model ( 2), the hypotheses related to prestige (H1) and age (H3) could also be partially accepted.In general terms, model (2) produces better results to explain the MOOC offering by universities.

Conclusions
The literature analyzed provides a broad perspective that reflects, among the most important aspects, the concern by researchers for the way in which universities must face the challenges posed by the new educational models.MOOCs constitute a new scenario in virtual education that has a key place in the academic debate on the present and future of universities.While far from being a solution to the weaknesses and barriers of the educational system, they present important challenges for educational communities in terms of rethinking, in all disciplines of knowledge, the teaching-learning processes within a context of a society that is undergoing permanent change and is shaped by the tension of the dynamics of globalization and the mass media.
This article provides empirical evidence of the relationships between the variables associated with the institutional profile of universities in different regions and their level of massive open course offerings, through the four most commonly used global platforms during the MOOC boom.Five hypotheses were tested, one associated with a strategic factor, the prestige of the university, and four associated with internal factors, namely the public or private nature of the university, the size, the age and the geographic region.Two external factors were considered as control variables in the analysis: GDP per capita and the Internet penetration in the university's country of origin.Given the importance the strategic factor "prestige" had in this study, it was estimated by two models, one using presence in the SHANGHAI ranking and the other using 4 categories according to the WEBOMETRICS ranking.
The results obtained in the regression analysis for both models suggest interesting data for the interpretation of the level of MOOC offer in relation to the ranking variables that measure the prestige of a university.When prestige is measured by the Shanghai ranking, this variable does not prove significant in order to explain the level of MOOC offering.However, prestige is a significantly influential variable when measured by the Webometrics ranking, in line with Ospina and Zorio (2016).
In the light of the results obtained, it must be kept in mind that the Webometrics ranking is based mainly on the web impact of the universities, in other words, their visibility, and not as much on the impact of their scientific activity, as in the Shanghai ranking (Aguillo et al., 2010;Aguillo et al., 2008;Docampo, 2008).It is therefore interesting to consider that those universities which have not necessarily earned their prestige only through mostly research work, but rather by building their visibility on the Web, are those that lead the MOOC offering, even though, as Daniel (2012) claims, they are not necessarily leaders in online instruction.This could be indicative of a trend in universities that are not precisely engaged intensively in research to excel by means of their MOOC offering, presenting themselves as the most innovative, an aspect also highlighted by the approach of prior research (see for instance Ospina & Zorio, 2016;Hollands & Tirthali, 2014).
Our study also reveals that private universities are the ones that have the most significant weight in the MOOC offering, in line with the studies by Hollands and Tirthali (2014) and Allen and Seaman (2014) based on North America.Our sample consists mainly of public universities (70%).However, it is the private universities that participate the most in the MOOC offer, as described by Shrivastava and Guiney (2014), who see greater resistance to this learning phenomenon in public institutions, not only due to the financial costs, but also the political costs that tend to be greater in public as opposed to private institutions (Gallego-Álvarez et al., 2011) and that arise particularly in relation to the teaching staff when the institution decides upon the financial model to support this type of innovation (Hollands & Tirthali, 2014).
In terms of the region, in spite of the fact that increasingly more and more European universities participate in the offer (Jansen et al., 2015) (in fact, the analyzed sample includes 55 European as compared to 60 North American institutions), the study shows, to a high level of statistical significance, that the North American universities have the most intensive MOOC offering.References to the magnitude of risk capital investments and donations by private organizations to support this innovation in the USA, in relation to both universities and other suppliers (Belleflamme & Jacqmin, 2014;Hollands & Tirthali, 2014), can explain this result, as could the results of Aguillo et al. (2008), whose analysis of the Web ranking revealed a digital academic gap between North American universities and their European counterparts.However, the results obtained when measuring prestige with the WEBOMETRICS variable also reflect the importance of Europe in this trend, which leads to the speculation that this region is increasingly accessing mechanisms available on the Web to improve their indicators of international visibility.
Moreover, the result from model (2), which measures prestige with the WEBOMETRICS variable, support the hypothesis that the newest universities participate more intensively in the MOOC offering, which suggests that there is a more cautious attitude on the part of the oldest universities towards this type of innovation, participating with one or only a few courses, which is more in line with the "wait and see" approach described by Hollands and Tirthali (2014).
Likewise, the results of measuring prestige with the SHANGHAI variable support the hypothesis related to size, as measured by the number of faculty members at the university.However, in the presence of the WEBOMETRICS variable, the number of faculty members cannot be identified as a significant factor of the number of MOOC offered.The explanation for this could lie in the fact that both in North America (e.g., University of Maryland, 2013; University of Illinois, 2014; West Virginia University, 2015) and in Europe (Jansen et al., 2015;EUA, 2015), many of the efforts related to this phenomenon are promoted by one-off institutional projects and not as an activity forming part of the teaching mission, which implies an increase in staff.
The results obtained suggest implications in the field of education, as they permit an empirical approach to the understanding of factors that have an influence on whether universities decide to participate more or less intensively in an initiative that in many ways is signaling new trends in how we conceive education.Likewise, it indicates relevant aspects for university policy on educational innovation, as universities must make strategic decisions related to their institutional philosophy, for example, whether or not to build their prestige based on greater international visibility, participating in the massiveness trend associated with x-MOOC-type platforms.Along these lines, there are still many challenges left to overcome, among which the financial aspect is one of the most significant.In relation to costs, a complex debate has arisen in relation to teaching staff, such as that described by Hollands and Tirthali (2014), that the first edition of a course is taught by associate professors and subsequent editions by non-tenured or outsourced instructors.Others consider the recognition of MOOCs within the European Credit Transfer and Accumulation System (ECTS) to be an opportunity (Schuwer et al., 2015).This possibility aside, it would be interesting to know whether the assessment of these courses by MOOC students' prospective employers depends on the prestige of the platform or of the university offering them (Rodriguez, 2012).In any case, the opportunity should be taken to develop a distinctive MOOC mission for the university or use it to improve its different missions, which as Daniel (2012) states, "would mean a true revolution in MOOCs" (pp.14).
Finally, one limitation of our study should be acknowledged.As MOOC offering is a growing trend that receives increasingly more financial backing from different organizations and agents, such as in the recent case of Europe (Jansen et al., 2015), the data are constantly changing.Nonetheless, future research may incorporate new variables , analyzing maybe the changes in the MOOC offering over time and understanding how universities around the world become involved in this global trend.It will be very interesting to find out how universities respond to the innovation needs of online instruction and whether they adopt different specialization strategies.
Este trabajo aporta evidencia empírica de las relaciones entre variables asociadas al perfil institucional de universidades de distintas regiones y su nivel de oferta de cursos masivos y abiertos, a través de las cuatro plataformas globales más difundidas en la etapa de auge de los MOOCs.Se contrastaron cinco hipótesis, una asociada a un factor estratégico como es el prestigio de la universidad y cuatro asociadas a factores internos como el carácter público o privado de la universidad, el tamaño, la antigüedad y la región geográfica.Dos factores externos fueron considerados en el análisis a modo de variables de control, el PIB per-cápita y la penetración de Internet en el país de origen de la universidad.Dada la importancia del factor estratégico «prestigio» para el estudio, se realizó la estimación en dos modelos, uno utilizando la variable SHANGHAI y el otro utilizando la variable WEBOMETRICS.
Los resultados obtenidos en la regresión para ambos modelos sugieren datos interesantes para interpretar el nivel de la oferta MOOC en relación con las variables de ranking que miden el prestigio de la universidad.Cuando el prestigio es medido por el ranking de Shanghai, no resulta significativa esta variable para explicar el nivel de oferta MOOC, a pesar de que los resultados descriptivos reflejan que la mayoría de las universidades están presentes en dicho ranking; mientras que sí resulta ser el prestigio una variable influyente de manera significativa cuando se mide con el ranking de Webometrics, acorde con los resultados de Ospina y Zorio (2016).

Table 1 .
Description of the sample

Table 2 .
Intangible Capital -http://dx.doi.org/10.3926/ic.79850th percentile, with 4, are 90 universities (58%), which demonstrates an important distance between the universities with the least and greatest offering.Descriptive statistics of continuous variables

Table 3 .
Frequencies of the categorical variables

Table 4 .
Correlations between continuous variables

Table 5 .
Contrast tests for categorical variables

Table 6 .
Multiple regression for factors that influence the MOOC offering