Research Questions and Study Framing
3.1. Research questions
The overarching goal of this study is to improve understanding of MOOC learning defined as development in the way that one interacts with others in forum discussions. Such development is indicated by changes in a learner’s participation pattern that are constructive with respect to substantive interaction about the course content. This study uses position as a conceptual construct for learner’s way of participation that can be characterized by characteristics related to their forum contributions and social relationships in the forum. Positioning is defined as the action of position taking. The research goal is addressed through answering the following research questions:
RQ1a: What common participation positions are found in MOOC forums? RQ1b: Which of the positions are found across different courses and different
time periods in the same course?
RQ2: What changes occurred in characteristics of individual learner’s position? RQ3: How did the changes manifest as common trajectories and represent
learning?
3.2. Study framing
3.2.1. Conceptualizing positions in MOOC discussions
Answering the research questions first calls for a conceptual framework that provides a holistic perspective on positions in MOOC discussions. Based on the literature review, the current study presents a two-dimension framework that
encompasses contribution characteristics and social characteristics which accounts for the nature of a learner’s contribution content and their social relationships respectively (see Table 3.1).
Table 3.1 Framework for position characteristics Dimensions Aspects
Contribution characteristics Related / unrelated to the course content Input seeking / input providing
Involving / not involving deep consideration of the discussion content Social characteristics Individual connectedness in the social network
Strength of social connections
For the first dimension in the framework, contribution characteristics are
assessed by three aspects about a contribution: does it (1) relate to the course content; (2) seek or provide input; and (3) contain deep consideration of the discussion content. First, MOOC discussions related and unrelated to the course content have been found to serve distinct purposes and differ in both interaction characteristics and social network properties (Wise, Cui, & Jin, 2017; Wise & Cui, 2018a). Therefore differentiating the two kinds of contributions is important for studying learner positions in the forum and MOOC learning. Content discussion involves interactions related to subject matter knowledge specified in the course syllabus and domain knowledge related to the course subject. Non-content discussion involves interactions not directly related to either form of knowledge, such as technical and logistical issues.
Second, input seeking and input providing activities differ in interaction purposes and the way they contribute to forum interactions. Learners who seek input request help or resource from others to meet their learning needs; at the same time their requests invite participation from others and potentially generate interaction opportunities (Liu et al., 2016; Stump et al., 2014). Learners who provide input contribute content and resources to forum discussion; their knowledge and expertise can be useful for
facilitating other’s learning, and is crucial for interactions in MOOC forums to sustain (Liu et al., 2016; Yang, Adamson, Rosé, 2014). Input seeking and input providing activities have been differentiated in MOOC studies that examined learners’ roles in discussion; input providers and input seekers have been found to differ in their learning outcomes measures by course grades (Hecking et al., 2017; Liu et al., 2016).
Finally, prior studies on MOOC and non-MOOC forum discussions revealed learners’ approaches to discussion content often associate with different levels of effort and learning that can be differentiated hierarchically (Anderson & Krathwohl, 2001; Wong et al., 2016). In this study the differentiation is made based on presence / absence of deep consideration of the discussion content. Deep consideration involves efforts to
understand and make sense of the content, such as explaining, elaborating, comparing, justifying, and reasoning whereas a non-deep contribution generally involves plain and straightforward information exchange through naming, listing, and describing content.
Social characteristics are assessed by two aspects: (1) individual connectedness in the social network and (2) strength of social connections with adjacent neighbors in the social network. Both aspects have been found useful for characterizing roles and understanding interactions in MOOC and non-MOOC contexts (Arnaboldi et al., 2017; Joksimovic et al., 2016; Poquet & Dawson, 2016; Wise & Cui, 2018a). In summary, the conceptual framework for position consists of two dimensions (five aspects): contribution characteristics (content-relatedness, input seeking / input providing, deep consideration) and social characteristics (connectedness in the social network, strength of social connections).
3.2.2. Important decisions in operationalizing the conceptual
framework to understand positions and position trajectories
From an empirical perspective, addressing the research goal of understanding MOOC learning by investigating positions and position trajectories involves multiple operational decisions. First, to investigate changes in a learner’s position during a MOOC, the course needs to be divided into multiple time segments. In this study a course is divided into three segments each with a similar number of weeks. This decision was made based on multiple considerations. For one thing, the level ofparticipation in MOOC forum is highly variant. Segmenting the course on a weekly basis can produce a lot of time segments without participation. Multi-week segments can alleviate this problem. Moreover, the three-segment solution allows for the potential for examining trajectories with multiple stages and the three segments respectively
correspond to the beginning, middle, and end phases in a course.
Second, to characterize learners’ positions in discussions related to the learning of course content, a participation profile needs to be built for each learner using their contribution characteristics and social characteristics. Contribution characteristics were extracted through performing content analysis on discussion contributions. As
documented in the literature review, a content analysis scheme for discussions in the MOOC context needs to be developed based on the contribution dimension in the
conceptual framework (see Table 3.1). Content-related discussions have been found to be more useful for understanding interaction about the course content and predicting learners’ course grades, in comparison to content-unrelated discussions and
undifferentiated discussions (Cui & Wise, 2015; Wise, Cui, Jin, & Vytasek, 2017; Wise & Cui, 2018a). Therefore this study extracts contribution characteristics from content- related discussions. Content-relatedness of a discussion is determined at thread level given that MOOC discussions are presented to learners as threaded conversations. Threads with a content-related initiating message are considered content-related as the initiating message largely sets the direction and scope of discussion. Considering that discussions with a non-content initiating message may change direction as new
participants join (Stump et al., 2013), for threads with a non-content initiating message, the decision is made based on whether or not the thread contains a substantial
proportion of content-related replies.
As for social characteristics, extracting these characteristics first requires constructing the social network. Prior research had indicated content and non-content discussions in MOOCs are participated by substantially different people (Wise & Cui, 2018a). Since this study aims to understand forum participation that occurs around the course content, the social networks are constructed for content-related discussions. Choosing a tie definition for constructing the social network is also an important decision. In MOOC forums, a learner can establish social connections through replying to other’s messages. They can also establish connections through accessing and being informed by multiple posts before making their own. Wise and Cui (2018a) presented the Limited Copresence tie definition to operationalize this assumption about social connections in MOOC social networks. Limited Copresence defines a tie as being present in the same part of a discussion based on the assumption that a participant in a thread with a small number of replies has ties with all others in the same thread. However the number of messages MOOC learners read before posting has not yet been empirically verified. Without setting a limit, the copresence assumption becomes problematic when size of thread is large and when many distinct people are involved in the same thread, which are often the case in MOOCs. An alternative tie definition is the Direct Reply definition that constructs a tie only if there is a direct reply relationship between two nodes in the same thread, without making any assumption about others who may have been
research. In comparison to the Limited Copresence definition, it results in social networks with similar overall structures but lower degree and edge weight (Fincham, Gaševic, & Pardo, 2018; Wise & Cui, 2018a). The current study uses the Direct Reply definition for constructing the social networks. The implications of this operational decision will be discussed in Chapter 6.
Third, to gain a big picture understanding of the common participation patterns in MOOC forums, this study needs to identify groups of learners that have the similar positions in discussion, thus participation profiles for learners in the same time period need to be grouped based on similarity in contribution and social characteristics. Cluster analysis is used for this task. Cluster analysis is an unsupervised machine learning method useful for discovering structures in data, and has been widely adopted for discovering groups of learners based on similarity in multiple types of participation characteristics in MOOC and non-MOOC forums (Eynon, Hjorth, Yasseri, & Gillani, 2016; Poquet & Dawson, 2016; Wise, Speer, Marbouti, & Hsiao, 2013).
Fourth, learners who participated in multiple time periods can be considered to have one position in each period; movement between these positions forms a learner’s position trajectory. Changes that occurred over time in learners’ positions are identified through comparing characteristics of the start position and the end position in the trajectory. Given that each position profile consists of multiple characteristics, for each identified position this study assigns a label that represents each characteristic, in addition to a profile name that highlights the key characteristics of the position. To understand common changes in position, the most frequent trajectories are examined both quantitatively through identifying the direction of changes and qualitatively through case studies.
Finally, to understand how variations in learning context may relate to position and positioning, two MOOCs offered in different subjects were selected for the study. To control for other differences in the learning context, the two selected courses were from the same course platform and had similar course length, number of forum participants, quantity of forum activities, and course policy for forum participation.