Absorptive Capacity and the Adoption of MOOCs in Higher Education… · 2016-06-21 · Massive Open...

Absorptive Capacity and MOOC Adoption

Thirty Sixth International Conference on Information Systems, Fort Worth 2015 1

Absorptive Capacity and the Adoption of MOOCs in Higher Education: The Role of

Educational IT Completed Research Paper

Peng Huang R.H. Smith School of Business

University of Maryland 4349 Van Munching Hall College Park, MD 20742

[email protected]

Henry C. Lucas, Jr. R.H. Smith School of Business

University of Maryland 4341 Van Munching Hall College Park, MD 20742 [email protected]

Abstract

Advanced information technologies have enabled the development of Massive Open Online Courses (MOOCs), which have the potential to transform higher education. Why are some schools able to more easily embrace this technology-based model of teaching, while others are reluctant to jump aboard? Applying the theory of absorptive capacity, we examine the role of a school’s educational IT – including its investments in both IT capabilities and IT governance structures – in becoming a MOOC producer. Using a unique longitudinal dataset that combines the complete history of MOOC adoption by US colleges and universities, and their use of educational IT, we find that prior educational IT capabilities such as 1) the use of Web 2.0, social media and other interactive tools for teaching, and 2) the school’s prior experience with distance education and hybrid teaching are positively associated with MOOC adoption. We also find that the contribution of educational IT capabilities to MOOC adoption is moderated by IT governance practices. When the provision of educational IT supporting services are highly decentralized, educational IT capabilities have a greater impact on the probability of a school adopting a MOOC than when these services are primarily provided by the central IT organization. We discuss the implications for research and practice.

Keywords: MOOC, online education, absorptive capacity, educational IT, IT governance

Introduction

Massive Open Online Courses (MOOCs) are both disrupting and transforming higher education (McAndrew and Scanlon 2013). Since the first MOOC appeared in October 2011, 123 American universities have offered these courses, with most of them on one of three major platforms – Coursera, EdX or Udacity. As of the time of this writing over 1300 MOOCs have been taught or announced.1 MOOCs have generated much interest and have enthusiastic supporters as well as determined opponents (Vardi 2012).

What exactly is a MOOC? The “massive” part of the title refers to the extraordinarily large number of students who take these courses. An early MOOC by Sebastian Thrun, a former Stanford Computer

1 The statistics are based on our analyses of data obtained from Class Central.



Science Professor and the founder of Udacity, drew 160,000 registrants. One of the authors of this paper has taught a MOOC on Coursera and found that students from over 150 countries registered for the course. “Open” means that there is no restriction in terms of registration and anyone with an Internet connection can take the course. In the beginning of the MOOC movement there was no charge for any of the courses; today one can pay for a certificate, but as yet there are few MOOCs offered for college credit. “Online” means that the courses are available on the Internet and the teaching model is enabled by a series of information and communication technologies. Online also implies that there is limited or no direct interaction between MOOC instructors and students as in a typical classroom setting. Instructors who choose to do so interact with students via announcements and the discussion boards. In his MOOC, one of the authors held weekly Google Hangouts with a small number of students; these Hangouts were streamed on YouTube and saved there for future viewing.

There are tremendous potential benefits associated with MOOCs for the universities that are experimenting with this new innovation. A recent report surveying 83 faculty, staff and administrators at 62 institutions identified six major goals for undertaking a MOOC (Hollands and Tirthali 2014). Among them are to extend the reach of the institution and provide greater access to education, and to maintain the school’s brand. There is also hope that MOOCs will eventually lower costs and/or increase revenues. However, most of the schools in the survey for now regard MOOCs as an investment rather than a way to make money. Officials at some schools feel that MOOCs will improve educational outcomes for participants and on campus students. Schools also view MOOCs as a form of innovation in teaching and learning, and wish to conduct research on teaching and student learning.

MOOCs represent a new challenge and opportunity for universities. A possible strategy to maximize opportunities and minimize threats is to explore the world of MOOCs, to become a content producer in preparation for the day when the business models associated with MOOCs mature (Dellarocas and Van Alstyne 2013) and schools are licensing MOOCs from various producers. However, it is unclear why some universities respond to this emerging innovation more swiftly than others, and theoretical works are lacking in this emerging field. More importantly, there is a surprising gap in the understanding of the role of educational Information Technology in influencing the adoption of MOOCs among universities. The purpose of this paper is to take the first step in identifying how a school’s prior use of educational IT, and its IT governance structure jointly shape the decision to adopt MOOCs. We use the theory of absorptive capacity (Cohen and Levinthal 1990, Roberts et al. 2012) to guide our empirical investigation of these questions. Using a unique longitudinal data set on the complete history of MOOC adoption by US colleges and universities, and their use of educational IT over a 3-year period, we find that such adoption decisions are significantly influenced by the schools’ prior use of educational IT, such as 1) their use of social media, web 2.0 technologies, or other interactive learning tools; and 2) their prior experience with e-learning and hybrid learning. Interestingly, we also find evidence that the effect of prior educational IT use is moderated by the schools’ IT governance structures: for example, prior educational IT use only has an effect on MOOC adoption when it is coupled with the decentralized provision of educational IT support services, and has no effect when supporting services are primarily provided by a central IT organization. These findings are consistent with the theory that the development of IT-related absorptive capacity depends not only on the stock of prior related knowledge, but also on complementary organizational capabilities (Jansen et al. 2005, Van Den Bosch et al. 1999). The results also provide evidence for the theory that complementary organizational assets play an important role in determining the value of IT investments (Brynjolfsson and Hitt 2000, Melville et al. 2004).

Studying MOOC adoption among higher education institutions has great practical implications. Many IT-enabled innovations are characterized by network effects (Katz and Shapiro 1994), particularly those platforms that enable two-sided or multi-sided markets (Parker and Van Alstyne 2005). In these markets the competitive outcomes often appear to be winner-takes-all, resulting in monopolies (Kemerer et al. 2013). The MOOC offerings are likely to be subject to the same competitive dynamics: for example, as the number of registrants for a course increases dramatically, it is easier for the learners to find classmates and join virtual discussion forums and local study groups, to access learning materials and study guides accumulated by past students on wikis and blogs, and to seek peer assessments, etc. On the supply side, MOOCs have great scale economy – as the number of students increases, the marginal cost of adding new students is almost negligible. Digitization of course content – video, audio, and e-learning materials – also lowers the marginal cost of offering a course in the future. Therefore, early adoption of MOOCs may afford universities an advantageous position in future competition in the higher education industry.



Identifying the early adopters of MOOCs should reveal which universities are most likely to benefit from this IT-enabled innovation.

Our paper contributes to prior research in several ways. First, we add to the literature on digital innovations (Fichman et al. 2014) by showing that IT capabilities play a key role in the adoption of such innovations, and we categorize and identify several core educational IT capabilities that are most influential in driving the adoption of MOOCs. To the best of our knowledge, our work is the first large-sample empirical study that formally investigates the adoption of MOOCs, a potentially disruptive pedagogical model that may transform higher education globally. Second, while the absorptive capacity framework (Cohen and Levinthal 1990) has been widely applied to examine firm strategy and behavior (Lane et al. 2001, Van Den Bosch et al. 1999), the applicability of this theory has rarely been tested in the context of non-profit organizations such as colleges and universities. We add new evidence of how the ability of universities to identify, assimilate, and apply IT-related external knowledge depends on their prior IT capabilities, and the fit between IT capabilities and IT governance. Third, we advance prior IT value literature that studies the role of IT complementarities (Brynjolfsson and Hitt 2000, Brynjolfsson and Milgrom 2012, Melville et al. 2004) and show how IT capabilities need to be coupled with organizational practices to enhance their value creation in the context of the higher education industry. Finally, a well-known difficulty in the study of IT value is the availability of micro-level IT investment data. In this study we also introduce a novel data set on IT investments made by a large sample of universities in the US – the Educause Core Data Service – that can be employed in future studies on the value of IT.

MOOCs as IT-enabled Innovation in Higher Education

Universities are conservative institutions that change slowly over many years. While the research and development carried out by the universities involve much creative thinking and innovative activities, on the teaching side universities do not have a long history of innovation. Modern classrooms have a bit more technology than 50 years ago, but in many classes there is a professor interacting with (often lecturing) a group of students. The advent of MOOCs presents a rare exception. Not only are instructor-student interaction, content delivery, grading and performance evaluation of MOOCs dramatically changed from a traditional classroom setting, MOOCs also differ from traditional education in terms of scale economies, network effects, and business models. In this work we conceptualize MOOCs as a form of IT-enabled innovation, consistent with Fichman et al. (2014)’s definition of “a product, process, or business model that is perceived as new, requires some significant changes on the part of adopters, and is embodied in or enabled by IT”. MOOCs as an innovation are both transformational and disruptive to the higher education industry.

How are MOOCs potentially transformational? First, they are highly scalable and offer the possibility for thousands of people who could never attend a major university to take a course offered by a highly regarded professor. Supporters of MOOCs view them as a way to raise educational levels around the world. Second, there are those who feel that MOOCs may be a first step at reducing college costs since so many students can access the work of a single faculty member; the marginal cost of adding one more student is very low. MOOCs may also change the way of accessing higher education: a possible scenario for the future is a college student mixing on campus classes from her university with MOOCs from other colleges, or taking the best-of-breed offerings from various MOOC-offering universities.

How are MOOCs potentially disruptive? First, they have some pedagogical issues that lead to questions about content delivery and learning outcomes assessment. As mentioned above, there is limited contact between students and the faculty, although some of the platforms are looking at ways to have local discussion leaders for courses. Grading is also an issue; courses that lend themselves to objective tests are well suited to online delivery. However, courses where essays are more appropriate for now rely on peer grading. On Coursera, for example, each student submitting a peer-graded assignment is assigned a number of others (typically four or five) to evaluate anonymously. The system awards the median grade of the graders to each student. Survey data also suggest that MOOCs suffer from high attrition rates and very low completion rates, possibly due to the lack of a social environment that facilitates sustained student engagement (Adamopoulos 2013, Yang et al. 2014).



Second, MOOCs are also potentially disruptive if schools use them to reduce prices. Georgia Tech is offering a MOOC-based MS in Computer Science for under $7,000 (Lewin 2013). It has received financial support from AT&T for this effort, so the tuition is partially subsidized. There are very few major universities that could offer an MS degree in any subject for that price and remain viable. MOOCs also have the potential to dramatically change the competitive dynamics for different types of schools. The small, private colleges are at the most risk. Will students apply and enroll in a school that is not well known or will they flock to institutions that offer MOOCs for credit taught by leading faculty across the world for a fraction of the cost?

Third, MOOCs are potentially disruptive because they may change the business model that colleges and universities have relied upon for many years. For example, MOOCs provide a model for start-up universities with much different business models than traditional ones, such as Project Minerva.2 More importantly, now schools will compete not at just the level of the university, but at the level of individual courses. There is also the fear that a few leading elite instructors will dominate a course topic, essentially resulting in a monopoly over a discipline’s content and collective singular thinking (Decker 2014). Lucas (2014) foresees a future business model that differentiates between the MOOC content producer and the content consumer. The content producer creates MOOCs and hopes to see them distributed to and offered by other schools for college credit in return for a royalty or shared revenue with the platform provider. The producer expands its brand globally and generates revenue in the process. The content consumer will be able to supplement courses with material from the top faculty in the world.

Theory and Hypotheses

Prior Educational IT Capability and MOOC Adoption

The theory of absorptive capacity (Cohen and Levinthal 1990) offers a unique framework in analyzing the relationship between prior educational IT capabilities and the adoption of IT-enabled innovation in teaching such as MOOCs. Under the context of organizational learning, absorptive capacity is defined as an organization’s ability to “recognize the value of new, external knowledge, assimilate it, and apply it to commercial ends” (Cohen and Levinthal 1990). Many point out that the development of absorptive capacity over time requires the accumulation of relevant knowledge base (Lane and Lubatkin 1998), and is usually path-dependent in that absorptive capacity depends on prior related knowledge (Roberts et al. 2012).

We maintain that there exists a positive relationship between a university’s prior educational IT capability and the likelihood of adopting MOOCs for two reasons. One, stronger IT capability helps accumulate business-IT knowledge, thereby enhancing an organization’s IT-related absorptive capacity (Roberts et al. 2012).Two, an expansive business-IT knowledge base is conducive to the identification, assimilation, and application of external knowledge related to new IT-enabled innovations. We elaborate on these two processes in some detail.

First, business-IT knowledge is referred to as “the combination of IT-related and business-related knowledge possessed by and exchanged among IT managers and business unit managers”, and it is an integral component of an organization’s overall absorptive capacity (Boynton et al. 1994, Nelson and Cooprider 1996). Universities where educational professionals enjoy high degrees of autonomy, such business-IT knowledge is likely to be possessed by two types of individuals: faculty members and supporting IT professionals. IT competence in faculty members includes IT-related explicit and implicit knowledge faculty members possess which enables them to pursue excellence in education. For example, frequent and repeated use of learning management systems, online assessment tools, or the use of social media and Web 2.0 tools for educational purposes by faculty members strengthens their understating of the use of IT to achieve effective content delivery and student evaluation. Similarly, business competence in IT professionals refers to “the set of business and personal knowledge and skills possessed by IT professionals that enable them to understand the business domain, speak the language of business, and interact with their business partners” (Bassellier and Benbasat 2004). Stronger educational IT capabilities

2 http://www.minervaproject.com/



at a university enhance an IT professional’s understanding of the role IT plays in promoting effective learning, and the deployment of technologies to achieve the strategic objectives of the university.

At the collective level, business-IT knowledge refers to the shared domain knowledge of faculty members and IT professionals, which consists of two dimensions – object knowledge and systems of knowing (Armstrong and Sambamurthy 1999). Systems of knowing represent a socialization capability that increases the group’s objective knowledge by facilitating the sharing of perspectives, pooling knowledge, and developing shared understanding (Nahapiet and Ghoshal 1998). Stronger prior educational IT capabilities also play a key role in creating shared domain knowledge. Universities with such capabilities are more likely to have developed structures of interaction among individuals and communication between faculty members and supporting IT professionals, leading to aligned business and IT objectives (Kearns and Sabherwal 2007).

Second, schools that have accumulated greater business-IT knowledge are more likely to recognize the opportunities presented by emerging, new innovations in teaching such as MOOCs, and are better positioned to assimilate and apply these innovations. For example, schools that have already experimented with various types of e-learning and hybrid learning may have accumulated important insights about the cost structure, the scalability, the format of interactions, and the limitations related to technology-enabled distance learning. When they are presented with the new IT-enabled innovations such as MOOCs, they are better able to assess the potential benefits of adoption, and take actions to seize these opportunities. In addition, it is well known that related knowledge and knowledge diversity lowers the knowledge barriers presented in adopting complex technology innovations. In the case of MOOCs, such barriers may appear a daunting obstacle for some schools if they have not acquired related prior business-IT knowledge. For example, unlike a traditional classroom setting, MOOCs attract a much larger audience and offer limited interaction between the instructors and students, and MOOC instructors have to find creative ways to engage students and prevent student attrition. Teaching a MOOC is mostly conducted in an asynchronous fashion, instead of face-to-face, synchronous communications. MOOCs are enabled by advanced information and communication technologies such as wikis, discussion boards, video tutorials and other Web 2.0 technologies, many of which are unfamiliar to college faculty. In addition, performance evaluation and learning outcome assessment are usually difficult in MOOCs, especially for non-technical subjects that involve writing and analysis. As a result, many MOOC courses have adopted technology-enabled peer grading systems where students in the class, while being assessed, also become assessors (Pappano 2012). Therefore, even experienced instructors and highly skilled supporting IT professionals may encounter difficulties adapting to this new form of teaching and have to experiment with various alternatives to achieve better outcomes. On the other hands, schools with greater accumulation of business-IT knowledge may be able to overcome these obstacles by relating the new innovation to their prior knowledge, and recombine their existing knowledge to acquire new competence through the processes of learning by using (Attewell 1992). Therefore, the depth of related business-IT knowledge helps assimilate and apply IT-enabled innovations in teaching. In summary, we propose:

Hypothesis 1: Universities with a higher level of prior educational IT capabilities are more likely to become MOOC adopters.

Decentralized Educational IT Support Services

Prior research on IT value has emphasized the role of complementary organizational resources (Brynjolfsson and Hitt 2000, Brynjolfsson and Milgrom 2012, Melville et al. 2004), which include decentralized organizational structure (Bresnahan et al. 2002, Brynjolfsson et al. 2002), skilled labor and workplace organization (Brynjolfsson et al. 2002), or innovative management practices (Bloom et al. 2012). One stream of research places particular emphasis on role of complementary organizational capabilities in creating synergy with IT capabilities to drive absorptive capacity (Roberts et al. 2012). Particularly, an organization’s ability to absorb valuable external knowledge not only depends on its prior related knowledge, but also on its investments in a set of structures and processes that facilitate knowledge absorption. Prior research has highlighted two types of organizational capabilities that may influence absorptive capacity: coordination capabilities and socialization capabilities (Jansen et al. 2005, Van Den Bosch et al. 1999). While coordination capabilities emphasize an organization’s ability to manage dependencies among its various activities (Malone and Crowston 1994), socialization capabilities stress its



ability to produce a shared ideology that offers organizational members an attractive identity as well as collective interpretation of reality (Van Den Bosch et al. 1999).

There are significant advantages and downsides associated with both centralized and decentralized decision authority with regard to the deployment IT. For example, Xue et al. (2011) argues that delegating the authority of decisions related to IT to business units may reap the benefits of quality and timeliness of decision making because the business units are best positioned to make swift and informed decisions in response to their idiosyncratic local needs and changing environment and opportunities (Anand and Mendelson 1997, Nault 1998). On the other hand, decentralized IT governance may also raise the issues of control because of agency problems – the objectives of business unit and the organization are not always perfectly aligned (Holmstrom and Milgrom 1991, Jensen and Meckling 1992). Ultimately, the choice of IT governance mode depends on the tradeoffs between these costs and benefits.

We argue that there is a complementarity between high educational IT capabilities and decentralized provision of educational IT support services in driving IT-related absorptive capacity in a university. As Sambamurthy and Zmud (1999) correctly point out, whether an organization pursues a decentralized locus for IT decision making is predicated on the extent to which line managers in the operating units possess the requisite business-IT knowledge. For example, in firms where line managers are not equipped with sufficient business-IT knowledge and lack the understanding of IT management practices, imposing decentralized IT decision rights may result in a poor fit and inferior performance (Boynton et al. 1994, Brown and Magill 1998). As we argued in the previous subsection, higher prior educational IT capability results in the accumulation of business-IT knowledge. In the context of a university, this business-IT knowledge is most likely to be distributed in each academic unit instead of central university IT, residing with the faculty members and their departmental supporting IT professionals. Academic units are highly autonomous entities and have idiosyncratic needs that are best addressed locally. IT staff deployed at academic unit level have a deeper understanding of these idiosyncratic needs, and have built social bonds with faculty members in their respective departments or schools through repeated interactions. In addition, the exchange and sharing of business-IT knowledge between faculty members and IT professionals is more effective if the IT supporting staff and instructors are collocated, especially for sharing tacit knowledge. Therefore, the decentralized distribution of business-IT knowledge requires a compatible IT governance structure – decentralized provision of educational IT support – that complements higher levels of educational IT capability and facilitates the absorption of new knowledge. In summary, we hypothesize that

Hypothesis 2: Decentralized provision of educational IT support services positively moderates the relationship between educational IT capability and MOOC adoption.

Data

We assembled a unique longitudinal data set of MOOC adoption among higher education institutions in the United States, their institutional characteristics, and their use of educational IT during academic years 2011-2012, 2012-2013, and 2013-2014. Our data consist of three major components and come from three separate data sources.

MOOC Adoption. We developed a web scripting tool and obtained the complete list of MOOCs that were ever offered on any MOOC platform by the end of spring semester in 2014 from a MOOC aggregation service provider, Class Central.3 Class Central is an aggregator of MOOC course listings that continually scans for MOOCs from all MOOC platform providers and places them in a central repository, starting from the introduction of the very first MOOC, Introduction to Artificial Intelligence, offered by Sebastian Thrun and Peter Norvig from Stanford University in October 2011. In addition to the title and instructors of each MOOC, Class Central also provides information such as a link to the actual course URL, a short description of the course, the subject of the course, the higher education institution through which the MOOC is sponsored, the provider (platform) on which the MOOC is offered, the length/duration of the course, the start date, the status (finished, self-paced, in-progress or future course) and average student feedback rating (if the course is finished).

3 Class Central is available at: https://www.class-central.com/



We present a summary of MOOC offerings during our sample period by platform providers, by sponsoring universities, and by subjects in Figures 1-3. We observe that Coursera is the dominant platform of choice for offering MOOCs: among the 1,122 MOOC offerings in our sample (that is, the course is offered by a US higher educational institution), 755 (or 67.3%) were offered on Coursera, with EdX being a remote second with 136 course offerings. In addition, while the subjects of the courses vary widely, and encompass the range from sciences, engineering, art, to social sciences, the singular most popular subject is computer science (235 out of 1,122), followed by Statistics & Data Analysis. In total, 123 universities have offered MOOCs during the 3-year period after the introduction of the first MOOC, and the heavy adopters tend to be nationally renowned private schools and flagship state universities, a fact which is consistent with prior survey that a major incentive of adopting MOOC is to maintain a school’s reputation and prestige (Hollands and Tirthali 2014).

Figure 1. MOOC Offerings by Platform Providers

Figure 2. MOOC Offerings by Universities, Top 15



Figure 3. MOOC Offerings by Subjects

Institutional Characteristics. We also obtain various institutional characteristics, enrollment/ completion/ graduation profiles, as well as student/ faculty information from Integrated Postsecondary Education Data System (IPEDS) data center.4 The IPEDS was established as the core postsecondary education data collection program for the National Center for Education Statistics, which conducts a system of surveys designed to collect data from all primary providers of postsecondary education. It is built around a series of interrelated surveys to collect institution-level data in such areas as enrollments, program completions, faculty, staff, and finances. The surveys target all primary providers of postsecondary education, which amount to 7,735 universities and colleges in the United States in its most recent survey. The data is widely used by many educational study organizations, such as College Board, Peterson's, and U.S. News & World Report to compile their publications.

Educational IT Capabilities. Our third source of data is the Educause Core Data Service (or CDS) survey, from which we derive our measures of the use of educational IT in US higher education institutions. Educause is a nonprofit association which focuses on “analysis, advocacy, community building, professional development, and knowledge creation to support the transformative role that IT can play in higher education.” The organization has drawn over 1,800 colleges and universities and over 300 corporations serving higher education IT as its members. The annual CDS survey is organized into a set of required modules that collect core IT information and optional modules that collect more details on specific IT domains, and is sent out to all higher education institution members every year to be completed. Participating members gain access to the data report on core metrics, and research reports and analyses published by Educause. The participating institutions often use CDS data for communicating the value if IT, benchmarking IT budgets and staffing, and comparing IT department structure and service delivery with peer schools.

Sample

The sample of our analyses is defined in the following way. We start with the universe of 7,735 schools surveyed by IPEDS, which by the definition of IPEDS, fall into one of the following categories: 1) Degree-granting, graduate with no undergraduate degrees; 2) Degree-granting, primarily baccalaureate or above; 3) Degree-granting, not primarily baccalaureate or above; 4) Degree-granting, associate's and certificates; 5) Nondegree-granting, above the baccalaureate; 6) Nondegree-granting, sub-baccalaureate; 7) Not

4 Data is available at: http://nces.ed.gov/ipeds/datacenter/



reported; 8) Not applicable. We find that among the 123 universities that have ever offered one or more MOOCs in any academic year by the end of our sample period, the vast majority of them (108, or 88%) belong to degree-granting, primarily baccalaureate or above category, while adopters in other categories are extremely rare and almost negligible. Therefore, we remove the schools in all other categories from our sample of analyses and focus on the schools that are degree-granting, primarily baccalaureate or above. Next, we examine the control of the institution, which can be either public, private not-for-profit, or private for-profit. Because none of the private for-profit schools offered any MOOC during our sample period, and because the for-profit schools use a very different accounting system than the not-for-profit organizations, we further remove private for-profit schools from our sample. The data set is then matched with Educause CDS survey database, and we retrieve the universities that fall into the intersection of the two data sources.

Our data is organized in a cross-sectional time series format. To maintain the data format consistent with the survey data from IPEDS and Educause CDS, the time series are coded as academic years (which we define as September 1st to August 31 of the next year), instead of calendar years. Because the very first MOOC offering started during academic year 2011-2012 (in October 2011), our sample period consists of 3 academic years: 2011-2012, 2012-2013, and 2013-2014. To allow for a causal interpretation, all the independent variables are lagged for one year, meaning that we use institutional characteristics and IT capabilities in academic year (t-1) to predict MOOC adoption in year t. In total, our final sample consists of 1405 university-year observations for 589 universities over a 3-year period.

Variables

Dependent Variable

The primary dependent variable we are interested in is the MOOC adoption decision by a university in a particular academic year. We match the data on MOOC offering history collected from Class Central with the universities in our sample, and create a binary indicator variable MOOC_adoptioni,t , which is set to 1 if university i offered any MOOC course during academic year t, and 0 otherwise. In some of the robustness tests, we also use an alternative, secondary definition of the dependent variable, which is the number of MOOC course offerings by university i in academic year t.

In Table 1 we present a summary of the number of universities that have adopted MOOCs by academic year. We observe a strong trend of growing popularity of MOOCs in higher education, but the overall level of adoption remains low: while there are only 6 universities that adopted MOOCs (or a 1.30% adoption rate) in academic year 2011-2012, the number grows to 35 (a 7.28% adoption rate) in year 2012-2013 and 57 (a 12.31% adoption rate) in year 2013-2014.

Table 1: Number of MOOC Adopters by Academic Year

Academic year Non-adopters Adopters Total

2011-2012 455 6 461

(98.70%) (1.30%) (100%)

2012-2013 446 35 481

(92.71%) (7.28%) (100%)

2013-2014 406 57 463

(87.69%) (12.31%) (100%)

Total 1307 98 1,405

(93.02%) (6.98%) (100%)

Independent Variables

Our independent variables of interest capture the use of advanced educational IT – a proxy for stronger IT capability in education, and the degree of decentralization of IT support services. We derive both the measures from Educause CDS survey data.



Educational IT capability. Module 3 of Educause CDS data is designed to collect data on educational technology services provided by universities. Specifically, a question in this module asks the survey respondent to indicate the use/status of a series of learning technologies or practices during the prior fiscal year. These technologies consist of 21 items that range from the use of web 2.0 tools (such as wikis and blogs), the use of social media (such as facebook and twitter), the practice of e-learning and hybrid learning, technology enabled teaching (such as simulation, clickers, collaboration tools and lecture capture), to the adoption of e-books and e-textbooks. The survey respondents indicate the status of the use of each technology as one of the following: 1) no discussion to date, 2) considered but not pursued, 3) experimenting/considering, 4) in planning, 5) deployed sparsely, or 6) deployed broadly. We create an ordinal categorical variable for each technology use, with the value 1 assigned to “no discussion to date” and value 6 assigned to “deployed broadly”.

To generate meaningful categorization of the educational IT capabilities and reduce the dimensionality of the data, we used exploratory factor analyses (EFA) to find the underlying factors associated with these measurement items. We perform an EFA using iterated principal factors method on the 21 measurement items. The scree plot of eigenvalues after the factor analysis is presented in Figure 1. Both Kaiser’s stopping rule (retaining factors with Eigenvalues greater than 1) and scree test suggest that there are three major underlying factors (Rencher 2003). The factor loadings, using orthogonal varimax rotation, are presented in Table 2, where we only retain the factor loadings greater than 0.4 (blanks represent abs(loading)<.4). By examining the measurement items associated with each factor, we find the first factor (IT1) is mainly associated with the use of web 2.0, social media, and other interactive tools for educational purposes. The second factor (IT2) is primarily associated with a university’s prior experience with distance education and hybrid teaching. The third factor (IT3) captures the adoption of e-books and e-textbooks in teaching. Therefore we use these 3 principle factors as the dimensions of measuring educational IT capability. Particularly, for each factor (IT1, IT2 or IT3) we take the average score of the measurement items that load heavily on the factor (items with loading>0.4) as the value of that variable.

Figure 4. Scree Plot after Factor Analysis

Decentralized Educational IT Support. Module 3 of the Educause CDS survey also poses a question with regard to the organizational unit that is primarily responsible for a series of educational technology support services. These support services include designated instructional technology center, instructional technologists assistance, faculty group training in the use of educational technology, support for learning management system, and special support services for distance education, to name a few. For each support service, the respondent indicates if it is: 1) primarily provided by central IT; 2) shared between central IT



and other admin or academic units; 3) primarily provided by other admin office; 4) primarily provided by academic department, school or college; 5) primarily provided by multi-campus system; or 6) not provided. We count the number of measurement items for which the associated service is provided primarily by academic department, school or college as opposed to a central IT organization. The count number (with a range of 0-14) is used to measure the degree of decentralization of educational IT support services.

Table 2: Factor Loadings of Educational IT Use

Factor1 Factor2 Factor3 Uniqueness

Blog 0.474 0.715

Collaboration tools 0.824

Distance learning – local instructor and remote students 0.878 0.259

Distance learning – remote instructor and local students 0.633 0.578

Document manage 0.823

E-learning 0.835 0.316

E-portfolios 0.856

E-books 0.857 0.200

E-textbooks 0.754 0.295

Facebook 0.478 0.772

Gaming 0.520 0.710

Hybrid courses 0.648 0.554

Information literacy 0.856

Interactive learning 0.428 0.667

Learning objects 0.443 0.648

Lecture capture 0.774

Mobile apps 0.472 0.735

Open content 0.498 0.728

Simulation 0.488 0.737

Twitter 0.607 0.619

Wiki 0.611 0.569

Control Variables

Institution Characteristics. We control for financial information related to a university such as average in-state tuition and out-of-state tuition for full time undergraduate students. We include as controls the total number of all undergraduate students and graduate students (including both full time and part time) who are enrolled. We also include control of institution (public or private) in our model. To control for the degree of strategic importance of IT to the universities, we also include a binary indictor of whether the institution’s strategic plan includes strategies and directions for IT, which is derived from Educause CDS survey.

Student Profile. We use a series of student characteristics as control variables. Admission rate is closely related to selectivity of students, and it is defined as the number of admissions made divided by number of applications received from first-time, degree seeking undergraduates for the academic year. Full time student retention rate, as well as 6-year completion rate of Bachelor’s degree (defined as # of completers within 150% of normal time / adjusted cohort for 4-year institutions), is also included. In addition, we



control for the percentage of full time, first-time undergraduate students that receive any form of financial aid.

Faculty Profile. We include two characteristics of the faculty that are likely to be associated with a university’s likelihood of adopting a MOOC. The first is the percentage of faculty members that are either tenured or on tenure track. IPEDS reports detailed statistics on faculty rank and tenure status, which gives total number of faculty members that are tenured, on tenure track, or not on tenure track/no tenure system. We retrieve the sum of faculty that are either tenured or on track, and divide this number by the total number of instructional faculty to get the percentage of faculty that are tenured or on the tenure track. The second control is related to faculty member compensation. It is likely that highly compensated professors are more capable at their jobs, and are more willing to experiment with new technology and incorporate new pedagogical methods. Because faculty members are hired under different contracts, we operationalize this variable as average salary of faculty on an equated 9-month contract.

School Resources. To the extent that MOOC offerings take considerable amount of investments of financial and human resources, we construct variables that capture the resource endowment for the sample universities. First, we measure financial resource of a university by the value of endowment asset per full-time equivalent student. Income generated from endowment assets is instrumental in maintaining academic excellence of many universities, and a declining endowment asset may lead to decaying facilities and hamper a university’s ability to provide instructional services. To derive this measure, we obtain the value of endowment assets at the end of the fiscal year, and divide it by the sum of reported full-time equivalent undergraduate enrollment and graduate enrollment of the university. Second, instructional faculty is the most valuable asset of any higher education institution, and is particular relevant in the MOOC adoption decisions as faculty members are ultimately responsible for the planning, design, and teaching of MOOCs. We use the average number of full time instructional faculty per full-time equivalent student to measure the human capital resource.

We present in Table 3 the summary statistics of our major dependent variable and independent variables. For brevity we omit the summary statistics of the control variables form this table.

Table 3: Summary Statistics

# Obs Mean Std. Dev. Min Max

MOOC adoption 1405 0.070 0.255 0 1

# of MOOCs offered 1405 0.462 2.754 0 54

IT1 – web 2.0 and social media 1405 4.113 0.934 1 5.7

IT2 – distance learning 1405 4.462 1.344 1 6

IT3 – adoption of E-books 1405 4.158 1.241 1 6

Decentralization (on scale 0-14) 1405 0.639 1.770 0 14

Results

Baseline Results

We use binary logistic models as a starting point to analyze how MOOC adoption decisions are shaped by educational IT capabilities and the decentralization of IT support services. The results of the logistic models are presented in Table 4. In all the columns the dependent variable is the binary indicator variable of whether a university i offered any MOOC courses in academic year t, while all the explanatory variables are lagged for one year. We also include a set of time period (academic year) fixed effects in all models. We use hetersedasticity robust standard errors clustered by universities where applicable. In column 1 of Table 4 we present the result from a baseline model of pooled logistic regression. To examine the potential complementarity between educational IT capability and decentralization of IT support services, in column 2 of Table 4 we also include interaction terms between these two measures.

While the pooled logistic models do not take advantage of the panel data structure of our setting, we explicitly control for unobserved university characteristics by using random effect panel data logistic



models, which further decompose the error term into an individual-specific component and a population component. We present the results in column 3 (for the main effects) and column 4 (for the interaction effects). We do not include the conditional logit (also known as the fixed-effects logit) models because of the well-known incidental parameter problems associated with these models (Hsiao 2003). The use of such models would result in dropping the majority of the observations from our sample, because for a large number of schools the dependent variable does not vary over the years (they never offered any MOOC).

Finally, an increasingly popular approach of estimating longitudinal binary response data that accounts for unobserved heterogeneity is the generalized estimating equations (GEE) method (Wooldridge 2002). With a population-averaged approach, the coefficients of GEE estimates describe how the population-averaged response rather than one individual’s response is conditioned on the covariates. We present the regression results of population-averaged panel GEE models with binomial distribution and logistic link function in column 5 (for the main effects) and column 6 (for the interaction effects) of Table 4.

Table 4: Main Results

(1) (2) (3) (4) (5) (6)

Binary Logistic Random Effects Logistic

Population-averaged Logistic

IT1 - web 2.0 and social media 0.545* 0.397 0.799** 0.653 0.598** 0.342

(0.293) (0.301) (0.338) (0.523) (0.271) (0.283)

IT2 – distance learning 0.362* 0.232 0.687*** 0.452 0.479** 0.147

(0.207) (0.218) (0.255) (0.400) (0.217) (0.233)

IT3 – adoption of E-books -0.259 -0.243 -0.270 -0.536 -0.144 -0.144

(0.173) (0.181) (0.184) (0.326) (0.114) (0.175)

decentralization 0.062 -4.333*** 0.115 -7.476*** 0.087 -3.782***

(0.076) (1.300) (0.096) (2.553) (0.068) (1.076)

IT1*decentralization 0.468** 0.686* 0.361**

(0.223) (0.356) (0.142)

IT2*decentralization 0.446*** 0.800** 0.416***

(0.156) (0.325) (0.145)

IT3*decentralization -0.046 0.025 -0.015

(0.119) (0.199) (0.107)

Constant -21.660*** -19.026*** -23.574*** -33.192*** -19.207*** -18.900***

(5.043) (4.732) (5.582) (9.668) (4.508) (4.437)

Year fixed effects yes yes yes yes yes yes

Pseudo R2 0.4866 0.5126 -- -- -- --

Observations 1,405 1,405 1,405 1,405 1,405 1,405

Number of schools 589 589 589 589 589 589

Note: Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Other control variables include: IT strategic plan, Faculty/student ratio, Endowment per student, Faculty salary, Tenure/Tenure track, Admission rate, Student retention rate, Graduation rate, Percentage of student receiving financial aid, In-state tuition, Out-of-state tuition, Undergraduate enrollment, Graduate enrollment, Private school.

We find partial support for Hypothesis 1 that educational IT capabilities are related to MOOC adoption in all the models across different model specifications. Particularly, the results from column 1, 3, and 5 (for the main effects) suggest that two out of the three measures of educational IT capabilities are significantly associated with the likelihood that the university will adopt MOOCs: the use of web 2.0, social media, and other interactive tools for educational purposes (IT1), and the school’s prior experience with distance



education and hybrid teaching (IT2). The results also suggest that the adoption of e-books and e-textbooks (IT3) is not associated with the probability of MOOC offering. The marginal effects of the estimated coefficients are quite substantial: for example, calculation based on the results from column 5 (the population-averaged panel logistic model) suggest that a unit increase in IT1 (which has a mean of 4.11 and range of 1-5.7) leads to a 0.75 percentage point increase in the probability of offering a MOOC (p<0.1) (in comparison, the overall adoption rate in our sample is 6.98%, see table 1). Similarly, a unit increase in IT2 (which has a mean of 4.46 and range of 1-6) leads to a 0.60 percentage point increase in the probability of offering a MOOC (p<0.1).

We also find partial support of Hypothesis 2 under the different model specifications, as the interaction terms of IT1 X IT decentralization, and IT2 X IT decentralization are both positive and significant in columns 2, 4, and 6. The results suggest that the contribution of educational IT on MOOC adoption is significantly moderated by IT decentralization. For example, the marginal effect calculations based on column 6 (the population-averaged panel logistic model) show that when IT decentralization is at a low level (10% quantile of the sample, or at IT decentralization =0), one unit increase in IT1 is associated with 1.24 percentage point (and not significant) increase in the probability of MOOC adoption. However, when IT decentralization is at a high level (90% quantile of the sample, or IT decentralization = 2), a unit increase in IT1 is associated with a 3.38 percentage point increase in the probability of MOOC adoption (p<0.01). The marginal effect of IT2 is similarly moderated by IT decentralization: when IT decentralization is at a low level (at IT decentralization =0), one unit increase in IT2 is associated with 0.53 percentage point (and not significant) increase in the probability of MOOC adoption. However, when IT decentralization is at a high level (at IT decentralization = 2), a unit increase in IT2 is associated with a 3.11 percentage point increase in the probability of MOOC adoption (p<0.01). Clearly, our regression results are consistent with the theory that stronger educational IT capabilities need to be coupled with flexible IT governance policy in driving IT-enabled innovations such as MOOCs.

Alternative Models and Measurements

We further probe the robustness of our findings by relaxing the assumptions of our empirical models and exploring alternative model specifications or different measures of our variables, and present the results of these robustness tests in Table 5. First, we test the validity of our findings by examining alterative assumptions about the distribution of the error term: instead of assuming a standard logistic distribution of the errors, we use a normal distribution of the errors and run the panel data random effect Probit models. The results are presented in column 1 (for the main effects) and column 2 (for the interaction effects) of Table 5. We find the estimates are very similar to those resulted from the panel data logistic models.

Second, we use hazard models as an alternative to analyze the role of educational IT and the decentralization of IT support services in determining the time to event – in this case the offering of MOOCs. Hazard models (also referred to as survival, duration, or event history model) are useful in our setting because they directly model time to event and do not depend on the normality assumption imposed in some other models that needs to be corrected. Hazard models also allow for the occurrence of multiple hazard events (e.g. under our context a university may offer MOOCs multiple times during the sample period), and provide an approach to address the incomplete observation of survival times when censoring occurs (Hosmer et al. 2008). Specifically, we chose the Cox proportional hazard model as our model specification. This model is a semi-parametric specification that makes no assumption of the functional form of the baseline hazard and assumes that covariates multiplicatively shift the baseline hazard function. We present the results from the Cox proportional hazard models in column 3 (for the main effects) and column 4 (for the interaction effects) of Table 5. Again we find our results are robust to this alternative specification.

Finally, we try a different measure of our dependent variable: instead of using a binary indicator of adoption, we use the actual number of MOOC offerings by university i in academic year t. Since this measure represents count data, we employ panel count data models and specify a negative binomial distribution with log link function, again estimating the model using population-averaged GEE techniques. We present the results in column 5 (for the main effects) and column 6 (for the interaction effects) of Table 5. We note that the results are consistent with what we find when using binary adoption indicators as the dependent variable. In general, we find robust evidences that supporting the hypothesis



that stronger educational IT (the use of web 2.0 and social media, or prior experience in distance learning) lead to a higher probability of adopting MOOCs, and the positive relationship is significantly stronger when they are coupled with higher level of decentralization of IT support services. However, our results consistently reveal that the adoption of e-books shows no effect in determining MOOC adoption.

Table 5: Alternative Models and Measures

(1) (2) (3) (4) (5) (6)

Random effects Probit Survival model Count model with

# of MOOCs as dependent

IT1 0.445** 0.359 0.424* 0.297 0.342** 0.073

(0.176) (0.280) (0.219) (0.231) (0.156) (0.162)

IT2 0.347*** 0.240 0.319* 0.210 0.414*** 0.138

(0.130) (0.213) (0.165) (0.172) (0.112) (0.126)

IT3 -0.152 -0.284 -0.179 -0.193 -0.005 0.071

(0.097) (0.174) (0.122) (0.136) (0.086) (0.096)

decentralization 0.060 -3.998*** 0.053 -3.393*** 0.103** -2.304***

(0.049) (1.372) (0.057) (0.931) (0.045) (0.716)

IT1*decentralization 0.370* 0.387** 0.430***

(0.193) (0.195) (0.156)

IT2*decentralization 0.430** 0.314*** 0.153*

(0.175) (0.113) (0.091)

IT3*decentralization 0.008 -0.016 -0.113*

(0.108) (0.082) (0.066)

Constant -12.164*** -17.648*** N/A N/A -17.879*** -18.549***

(2.896) (5.212) (2.997) (3.094)

Year fixed effects yes yes N/A N/A yes yes

Observations 1,405 1,405 1,405 1,405 1,405 1,405

Number of schools 589 589 589 589 589 589

Note: Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Other control variables include: IT strategic plan, Faculty/student ratio, Endowment per student, Faculty salary, Tenure/Tenure track, Admission rate, Student retention rate, Graduation rate, Percentage of student receiving financial aid, In-state tuition, Out-of-state tuition, Undergraduate enrollment, Graduate enrollment, Private school.

Conclusions and Discussion

In this study we have constructed a novel data set from several sources to examine how prior use of educational IT and IT governance structure jointly influence universities’ decisions on adopting the technological innovation of Massive Open Online Courses. We conceptualize the production of MOOCs by a university as an IT-enabled innovation whose adoption requires universities’ ability to identify, assimilate, and apply external knowledge. Our results are consistent with the theory of absorptive capacity (Cohen and Levinthal 1990). They show that such adoption decisions are indeed path-dependent – schools that have developed stronger prior educational IT capabilities are more likely to become a MOOC creator. Particularly, among a wide range of educational technologies that we examine, 1) the use of social media, web 2.0 technologies and other interactively learning tools, and 2) a school’s prior experience with e-learning and hybrid learning are positively associated with the likelihood that it will offer MOOCs. In addition, we find that IT governance structure plays an important moderating role. For example, the positive effect of educational IT capabilities only has an effect when it is also coupled with decentralized provision of IT supporting services. When these services are primarily provided by central IT, the effect of educational IT capabilities diminishes. This result highlights the role of complementary organizational



capabilities in creating synergies with IT investments in shaping absorptive capacity (Jansen et al. 2005). Our research takes the first step in understanding how IT-related factors – including investments made by universities in creating both IT knowledge and IT structure – determine their reaction to emerging technology innovations that may potentially disrupt and transform higher education.

Theoretically, our work makes several useful contributions. First, we extend prior research on absorptive capacity that almost exclusively examines how it influences firm strategy and performance, such as its role in organizational learning through strategic alliances (Lane and Lubatkin 1998) or driving innovation (Tsai 2001). By applying this theory to a non-profit context such as higher education, we show the applicability of the theory is not limited to studying firm behaviors – the absorptive capacity of non-profit organizations also appears to be path-dependent and requires the accumulation of prior related knowledge.

More importantly, our work also echoes earlier studies that stress the role of complementary organizational capabilities in developing absorptive capacity (Jansen et al. 2005, Van Den Bosch et al. 1999), and we find decentralized IT support service deployment is one of such complementary capabilities in driving IT-related absorptive capacity. Our interpretation is that in a highly decentralized, autonomous organization with less hierarchical control, decentralized IT governance forms the basis of both coordination capabilities (Malone and Crowston 1994) that enhance the management of the dependence among faculty and IT professional. Decentralized IT brings together different sources of expertise, and socialization capabilities (Van Den Bosch et al. 1999) that offer organizational members a consistent set of beliefs and produces shared ideology. In this way, decentralized IT governance enhances knowledge exchanges within academic units and amplifies the effect of prior knowledge.

Finally, our work also contributes to the recent IT value research that emphasizes the role of complementary organization assets as a moderator that enhances the return of IT investments (Brynjolfsson and Hitt 2000, Brynjolfsson and Milgrom 2012, Melville et al. 2004). While most of the studies in this research stream have focused on firm productivity (typically measured by an production function that links input and output), we present evidence from an alternative perspective – organizational assets that are complementary to IT investment also play a role in driving innovation adoption.

Our work also provides practical implications for education professional. We have argued that MOOCs are a potential transformational and disruptive technology for universities. They are a part of a broader set of technologies that universities can use to create blended and flipped classes as well as online classes that feature synchronous interactive sessions with faculty and students. However, there is a fundamental difference between MOOC and other IT-enabled distance learning: the network effects and scale economics associated with MOOCs are greatly amplified, which imply that early entries will establish a huge first-mover advantage. Our results suggest that not all schools are on an equal footing – the adoption of MOOC is path-dependent and strong educational IT capabilities cannot be developed over night. The first-mover advantages are most likely to be harvested by schools that already possess such IT capabilities. They are most likely to become disruptors that threaten the survival of schools that remain solely as content consumers. In addition, schools that want to embrace the emerging MOOC innovation need to prioritize their IT capability building: among a broad set of educational technologies, the use of social media, web 2.0 tools, and other interactive learning tools should receive greater attention. They also need to experiment with various types of e-learning and hybrid learning to acquire related knowledge. Furthermore, they should acquire organizational capabilities that are compatible with advanced IT capabilities and build structures that facilitate knowledge exchange in a decentralized fashion.

As MOOCs become an increasingly popular supplement to traditional forms of education, our research can be extended in a number of ways. For example, while many hope that MOOCs will enhance learning outcomes by creating a personalized experience or motivating instructors to rethink their pedagogy (Hollands and Tirthali 2014), very little is known about their actual impact on educational outcomes. Systematic investigations that compare different formats of instruction are needed to assess the merits of MOOC, classroom teaching, and blended teaching. Another research direction is to investigate what types of courses are most conducive to MOOC pedagogy. For example, our data reveal that computer science is the single most popular subject among all the MOOCs have been offered by the universities in the US, but it is unclear if it is due to the nature of the subject, or because computer science professors are more comfortable using advanced technologies for teaching. In addition, social sciences such as humanities,



business & management, or history & philosophy are also among the popular subjects in our sample. It is unclear to what extent the MOOC model limits or enhances the teaching of a particular academic discipline or subject, and how various technologies can be employed to address these limitations. Furthermore, it is well known that many MOOC offerings suffer from high attrition rate and very low completion rate. Extant literature provides very little insight on the measures that MOOC instructors can take to address these issues. We call for future research to explore these interesting and important questions.

References

Adamopoulos, P. 2013. "What Makes a Great MOOC? An Interdisciplinary Analysis of Student Retention in Online Courses," in Proceedings of the 34th International Conference on Information Systems: Milano, Italy.

Anand, K. S., and Mendelson, H. 1997. "Information and organization for horizontal multimarket coordination," Management Science (43:12), pp 1609-1627.

Armstrong, C. P., and Sambamurthy, V. 1999. "Information technology assimilation in firms: The influence of senior leadership and IT infrastructures," Information Systems Research (10:4), pp 304-327.

Attewell, P. 1992. "Technology diffusion and organizational learning: The case of business computing," Organization Science (3:1), pp 1-19.

Bassellier, G., and Benbasat, I. 2004. "Business competence of information technology professionals: conceptual development and influence on IT-business partnerships," MIS quarterly (28:4), pp 673-694.

Bloom, N., Sadun, R., and Van Reenen, J. 2012. "Americans do IT better: US multinationals and the productivity miracle," The American Economic Review (102:1), pp 167-201.

Boynton, A. C., Zmud, R. W., and Jacobs, G. C. 1994. "The influence of IT management practice on IT use in large organizations," MIS Quarterly (18:3), pp 299-318.

Bresnahan, T. F., Brynjolfsson, E., and Hitt, L. M. 2002. "Information technology, workplace organization and the demand for skilled labor: Firm-level evidence," Quarterly Journal of Economics (117:1), pp 339-376.

Brown, C. V., and Magill, S. L. 1998. "Reconceptualizing the context-design issue for the information systems function," Organization Science (9:2), pp 176-194.

Brynjolfsson, E., and Hitt, L. M. 2000. "Beyond computation: Information technology, organizational transformation and business performance," The Journal of Economic Perspectives (14:4), pp 23-48.

Brynjolfsson, E., Hitt, L. M., and Yang, S. 2002. "Intangible assets: Computers and organizational capital," Brookings Papers on Economic Activity (2002:1), pp 137-181.

Brynjolfsson, E., and Milgrom, P. 2012. "Complementarity in organizations," in The Handbook of Organizational Economics, R. Gibbons and J. Roberts (eds.), Princeton University Press.

Cohen, W. M., and Levinthal, D. A. 1990. "Absorptive capacity: a new perspective on learning and innovation," Administrative Science Quarterly (35:1), pp 128-152.

Decker, G. L. 2014. "MOOCology 1.0," in Invasion of the MOOCs: The promises and perils of massive open online courses, S. D. Krause and C. Lowe (eds.), Parlor Press: Anderson, SC, pp. 3-13.

Dellarocas, C., and Van Alstyne, M. 2013. "Money models for MOOCs," Communications of the ACM (56:8), pp 25-28.

Fichman, R. G., Dos Santos, B. L., and Zheng, Z. 2014. "Digital innovation as a fundamental and powerful concept in the information systems curriculum," MIS Quarterly (38:2), pp 329-A315.

Hollands, F. M., and Tirthali, D. 2014. "MOOCs: expectations and reality. Full Report.," Center for Benefit-Cost Studies of Education, Teachers College, Columbia University.

Holmstrom, B., and Milgrom, P. 1991. "Multitask principal-agent analyses: Incentive contracts, asset ownership, and job design," Journal of Law, Economics, & Organization (7:sp), pp 24-52.

Hosmer, D. W., Lemeshow, S., and May, S. 2008. Applied Survival Analysis: Regression Modeling of Time to Event Data, (2nd ed.) Wiley.

Hsiao, C. 2003. Analysis of Panel Data, (2nd ed.) Cambridge University Press: Cambridge, UK.



Jansen, J. J., Van Den Bosch, F. A., and Volberda, H. W. 2005. "Managing potential and realized absorptive capacity: how do organizational antecedents matter?," Academy of Management Journal (48:6), pp 999-1015.

Jensen, M. C., and Meckling, W. H. 1992. "Specific and general knowledge and organizational structure," in Contract Economics, L. Werin and H. Wijkander (eds.), Blackwell: Oxford, pp. 251–274.

Katz, M. L., and Shapiro, C. 1994. "Systems competition and network effects," The Journal of Economic Perspectives (8:2), pp 93-115.

Kearns, G. S., and Sabherwal, R. 2007. "Strategic alignment between business and information technology: a knowledge-based view of behaviors, outcome, and consequences," Journal of Management Information Systems (23:3), pp 129-162.

Kemerer, C. F., Liu, C. Z., and Smith, M. D. 2013. "Strategies for tomorrow's' winners-take-some'digital goods markets," Communications of the ACM (56:5), pp 76-82.

Lane, P. J., and Lubatkin, M. 1998. "Relative absorptive capacity and interorganizational learning," Strategic Management Journal (19:5), pp 461-477.

Lane, P. J., Salk, J. E., and Lyles, M. A. 2001. "Absorptive capacity, learning, and performance in international joint ventures," Strategic Management Journal (22:12), pp 1139-1161.

Lewin, T. 2013. "Master’s degree is new frontier of study online," New York Times (available at http://www.nytimes.com/2013/08/18/education/masters-degree-is-new-frontier-of-study-online.html).

Lucas, H. 2014. "Disrupting and transforming the university," Communications of the ACM (57:10), pp 32-35.

Malone, T. W., and Crowston, K. 1994. "The interdisciplinary study of coordination," ACM Computing Surveys (CSUR) (26:1), pp 87-119.

McAndrew, P., and Scanlon, E. 2013. "Open learning at a distance: Lessons for struggling MOOCs," Science (342:6165), pp 1450-1451.

Melville, N., Kraemer, K., and Gurbaxani, V. 2004. "Review: Information technology and organizational performance: An integrative model of IT business value," MIS Quarterly (28:2), pp 283-322.

Nahapiet, J., and Ghoshal, S. 1998. "Social capital, intellectual capital, and the organizational advantage," Academy of Management Review (23:2), pp 242-266.

Nault, B. R. 1998. "Information technology and organization design: Locating decisions and information," Management Science (44:10), pp 1321-1335.

Nelson, K. M., and Cooprider, J. G. 1996. "The contribution of shared knowledge to IS group performance," MIS quarterly (20:4), pp 409-432.

Pappano, L. 2012. "The Year of the MOOC," The New York Times Nov. 2, 2012 (available at http://www.nytimes.com/2012/11/04/education/edlife/massive-open-online-courses-are-multiplying-at-a-rapid-pace.html).

Parker, G. G., and Van Alstyne, M. W. 2005. "Two-sided network effects: A theory of information product design," Management Science (51:10), pp 1494-1504.

Rencher, A. C. 2003. Methods of Multivariate Analysis, (2nd ed.) John Wiley & Sons. Roberts, N., Galluch, P. S., Dinger, M., and Grover, V. 2012. "Absorptive Capacity and Information

Systems Research: Review, Synthesis, and Directions for Future Research," MIS quarterly (36:2), pp 625-648.

Sambamurthy, V., and Zmud, R. W. 1999. "Arrangements for information technology governance: A theory of multiple contingencies," MIS quarterly (23:2), pp 261-290.

Tsai, W. 2001. "Knowledge transfer in intraorganizational networks: Effects of network position and absorptive capacity on business unit innovation and performance," Academy of Management Journal (44:5), pp 996-1004.

Van Den Bosch, F. A., Volberda, H. W., and De Boer, M. 1999. "Coevolution of firm absorptive capacity and knowledge environment: Organizational forms and combinative capabilities," Organization Science (10:5), pp 551-568.

Vardi, M. Y. 2012. "Will MOOCs destroy academia?," Commun. ACM (55:11), p 5. Wooldridge, J. M. 2002. Econometric Analysis of Cross Section and Panel Data, (1st ed.) MIT Press:

Cambridge, MA. Xue, L., Ray, G., and Gu, B. 2011. "Environmental uncertainty and IT infrastructure governance: A

curvilinear relationship," Information Systems Research (22:2), pp 389-399. Yang, D., Sinha, T., Adamson, D., and Rose, C. P. 2014. "“Turn on, Tune in, Drop out”: Anticipating

student dropouts in Massive Open Online Courses," Working paper, Carnegie Mellon University.

Date post:	23-May-2020
Category:	Documents
Upload:	others
View:	6 times
Download:	0 times

Absorptive Capacity and the Adoption of MOOCs in Higher Education… · 2016-06-21 · Massive Open...

Documents