Inclusiveness in open online computing education: geo-cultural perspectives

我的博士研究旨在研究人口统计学和社会经济地位在在线学习中的作用。我试图帮助我们理解在线学习者如何学习、参与和感知各种学习设计元素（例如，教学视频、阅读材料、基于讨论的活动和评估）。我的博士研究是在新冠疫情前进行的，重点是大规模开放在线课程（MOOCs）。尽管在线学习和教学社区强烈期望免费和广泛宣传的mooc可能会解决全球教育差距问题，但大多数活跃的学习者来自特定的发达国家。先前的研究表明，在线学习者在实现学习目标方面的成功程度随地理文化和社会经济维度以及学习设计特征而变化。尽管注册人数各不相同，但大多数mooc采用了一种“一刀切”的设计，向所有学习者提供相同的学习活动集和顺序。我的研究旨在研究如何在各种情况下大规模地调整学习设计，以提高学习者的持久性。该研究得益于一系列理论框架（例如，Extended-GLOBE和Hofstede ncd，文化适应性用户界面设计，其结构源自技术接受模型或TAM）。对于社会经济学的概念化，我使用了每个国家的国民总收入（GNI）。分析方法包括生存分析中的决策树、序列挖掘和交叉验证交互。混合方法研究采用半结构化访谈和人工中介问题来调查MOOC学习者对各种学习设计元素的感知的语境差异。该分析将定性（主题分析）方法与情感挖掘相结合。我的博士研究清楚地表明，与亚组/交互分析相比，在线学习数据的整体分析可以掩盖学习设计因素与学习者持久性之间相关性的地理文化和社会经济异质性。因此，总体数据分析结果主要反映了最大的子群体（例如，盎格鲁-撒克逊地缘文化群体，高收入国家）的行为模式，这可以与其他较小的子群体（例如，非洲或南亚，中低收入和低收入国家）的模式形成对比。因此，它可以改善大多数群体的结果，同时将代表性不足的群体的成员抛在后面。例如，我博士项目中的研究3使用了来自10个大型mooc的大量日志数据，研究了如何量化学习设计元素（例如，视频数量、阅读材料、基于讨论的学习活动和测验）与学习者持久性之间的预测联系，并在10个地理文化背景下有所不同。在定性研究中，研究4采用半结构化访谈来探讨学习者的认知。我发现，跨文化学习设计偏好可能因学科而异（例如，计算机、艺术和人文学科）。这项研究已经在解决部分拼图问题和勾画未来研究的新方向方面做出了有价值的贡献。在新冠肺炎疫情后的世界重新审视我的研究，真是再及时不过了。如果在更具体的上下文中执行复制，那么从我的博士研究中获得的启示和经验教训将显得更加相关。计算技术是MOOC最主要、最多样化的学科领域。因此，得益于25%的新研究津贴，我建议在在线计算教育的背景下，更具体地说，在树莓派基金会开发的计算MOOC的背景下，通过FutureLearn平台提供，基于大规模调查的混合方法复制我的研究3和4。伦理批准将在奖学金开始之前获得。

My PhD research set out to study the role of demographics and socioeconomic status in online learning. I sought to contribute to our understanding of how online learners learn from, engage with, and perceive various learning design elements (e.g., instructional videos, reading material, discussion-based activities, and assessments). My doctoral research was conducted in pre-Covid times and focused particularly on massive, open, online courses (MOOCs). Despite the strong expectations of the online learning and teaching community that free and widely advertised MOOCs may potentially address the global disparity in education, most active learners originate from specific developed countries. Prior work suggests that how successful online learners are in achieving their learning goals varies along geo-cultural and socioeconomic dimensions as well as with learning design features. But despite diverse enrolments, most MOOCs adopt a one-size-fits-all design that presents the same set and sequence of learning activities to all learners. My research set out to study how learning designs could be adapted at scale in various contexts to improve learners' persistence. The research benefited from a range of theoretical frameworks (e.g., Extended-GLOBE and Hofstede NCDs, culturally-adaptive user interface designs whose constructs are derived from the Technology Acceptance Model or TAM). For conceptualization of socioeconomics, I used gross national income (GNI) for each country. The analysis methods included decision trees, sequence mining, and cross-validated interactions in survival analysis. The mixed-method research used semi-structured interviews and artefact-mediated questions to investigate the contextual differences in MOOC learners' perceptions about various learning design elements. The analysis combined a qualitative (thematic analysis) method with sentiment mining. My doctoral research clearly demonstrated that in comparison to subgroup/interaction analyses, an overall analysis of online learning data could mask geo-cultural and socioeconomic heterogeneity in the correlations between learning design factors and learner persistence. Consequently, overarching data analysis results primarily reflect the behavioural patterns of the largest subgroup (e.g., Anglo-Saxon geo-cultural group, high-income countries), which can stand in contrast to patterns of other, smaller subgroups (e.g., African or South Asian, lower-middle- and low-income countries). As a result, it can lead to improved outcomes for the majority group while leaving behind members of underrepresented groups. For example, using the voluminous log data from ten large MOOCs, Study 3 in my PhD project examined how the quantified the predictive link between learning design elements (e.g., number of videos, reading material, discussion-based learning activities, and quizzes) and learners' persistence varies across the ten geo-cultural contexts. While the qualitative study, Study 4 used semi-structured interviews to explore learners' perceptions. I identified that cross-cultural learning design preferences may vary across the disciplines (e.g., computing, arts and humanities). This research has already made a valuable contribution in solving part of the jigsaw and outlining new directions for future research. Revisiting my research in a post-Covid world, could not be more timely. The implications and lessons learned from my PhD research will appear more relevant if a replication is performed in a more specific context. Computing technology is the most dominant and diverse MOOC subject area. Therefore, benefitting from the allowance of 25% new research, I propose large-scale survey based mixed-method replication of my studies 3 and 4 in the context of online computing education, more specifically, within the context of Computing MOOC developed by Raspberry Pi foundation, offered via FutureLearn platform. Ethical approvals will be attained well before the start of the fellowship.