PaperlayerView exact revision on arXiv

Source-linked AI summary

Virtual Global Collaboration in Data Analytics and Machine Learning Education: A Mixed-Methods Study of Asynchronous Cross-Border Teamwork

Sehrish Basir Nizamani, Saad Nizamani, Khyati Goyal, Sarwat Nizamani, Zannah Zeiw

arXiv:2610.01147v1cs.CY

TL;DR

Evidence is limited on short-term, structured virtual collaboration in computing that combines analytics and machine learning, particularly in asynchronous settings. This study analyzed survey and reflection data from a cross-border collaboration; all seven hypotheses were supported, and cross-cultural communication emerged as the unique driver of global readiness.

  • Problem

    Research is limited on short-term, structured virtual collaboration in computing that combines data analytics and machine learning, especially in predominantly asynchronous settings.

    Source: §I · p. 1, D. Summary and Research Gap · p. 2

  • Method

    The study analyzed Likert-scale construct responses using nonparametric tests and regression, alongside open-ended responses coded inductively.

    Source: E. Data Analysis · p. 4

  • Results

    All seven hypotheses were supported across U.S. and Pakistan cohorts; cross-cultural communication emerged as the unique driver of global citizenship and readiness, and mediators explained roughly 45 to 55% of outcome variance across models.

    Source: §VI · p. 6, §VIII · p. 7

  • Takeaways & Limitations

    The findings present short-term, structured VGC as a model for supporting technical learning, collaborative skills, and global readiness in undergraduate computing education.

    Source: §VIII · p. 7

    Self-reported measures from one short activity limit causal claims, and optional participation may favor students already motivated toward global collaboration.

    Source: §VII · p. 7

Abstract

from arXiv · show

This innovative practice full paper examines a structured Virtual Global Collaboration (VGC) activity between undergraduate computing courses in the United States and Pakistan. The project was designed to support technical and intercultural skill development through asynchronous international teamwork in data analytics and introductory machine learning. Students worked in mixed-institution teams through a six-phase project including cultural orientation, dataset selection, data cleaning, analysis, introductory machine learning, and structured reporting. Using shared computational tools, teams coordinated across time zones to complete a joint data-driven project. Survey and qualitative reflection data were analyzed to examine collaborative experience, communication and cultural dynamics, learning outcomes, perceived value, and global readiness. Results show consistently positive student experiences across both cohorts, with collaborative processes strongly associated with learning and perceived value, and cross-cultural communication emerging as the primary driver of global readiness. These findings demonstrate how short-term, structured VGC can be effectively integrated into computing courses to support both technical learning and global competence.

I. INTRODUCTION

Virtual global collaboration (VGC) brings international, discipline-specific teamwork into higher education through online projects and structured interaction. This study examines asynchronous cross-border teamwork in U.S. and Pakistani computing courses against prior findings on learning and intercultural competence.

I. INTRODUCTION

  • VGC connects students with international peers through course-embedded projects, enabling collaborative learning without traditional study-abroad barriers.Structured interaction and reflection can support intercultural competence, teamwork, negotiation, and problem-solving in distributed settings.Source: §I · p. 1
  • The study investigates how students experience and learn through primarily asynchronous U.S.–Pakistan teamwork, and how it shapes perceptions of collaboration, learning, and global readiness.Source: §I · p. 1

A. Learning Outcomes in Virtual Global Collaboration

  • Prior work reports that virtual collaboration supports course-concept understanding and professional skills, while one COIL study found intercultural gains only at one of two sites and on one of two measures.The second site and measure showed no significant difference.Source: A. Learning Outcomes in Virtual Global… · p. 1

B. Intercultural Competence and Global Awareness

C. Design Considerations and Challenges

Prior work identifies coordination challenges in virtual collaboration and emphasizes structured support. The paper addresses a computing-education research gap with a framework linking VGC processes to learning, perceived value, global readiness, and reflection on challenges.

C. Design Considerations and Challenges

  • Time zones, communication barriers, uneven participation, and technology can hinder coordination; structured tasks, clear expectations, and guided collaboration can mitigate these challenges.Source: C. Design Considerations and Challenges · p. 2

D. Summary and Research Gap

  • Research is limited on short-term, structured virtual collaboration in computing that combines data analytics and machine learning, especially in predominantly asynchronous settings.Source: D. Summary and Research Gap · p. 2

III. CONCEPTUAL FRAMEWORK

  • The framework treats VGC teamwork in analytics and machine learning as the instructional context, with collaborative experience, communication and cultural dynamics, and logistics as interconnected mediators.It anticipates effects on learning and skills, perceived value, global citizenship and readiness, and whether students view challenges as meaningful learning.Source: §III · p. 2
  • The proposed hypotheses predict positive collaboration, effective cross-cultural communication, manageable logistics, and positive mediator relationships with learning, perceived value, and global readiness.Source: §III · p. 3
  • H7 proposes that students will perceive collaboration’s opportunities and growth as outweighing its challenges.Source: §III · p. 3

IV. STUDY CONTEXT

The activity was embedded in undergraduate computing courses at U.S. and Pakistani public universities. Students voluntarily joined mixed-institution teams for data analytics and introductory machine-learning work.

IV. STUDY CONTEXT

  • The study took place in undergraduate computing courses at a large U.S. public research university partnered with a public university in Pakistan.Source: §IV · p. 3
  • The VGC activity was an optional component of courses focused on data analytics and introductory machine learning.Source: §IV · p. 3

A. Participants

  • Students from both institutions voluntarily opted in and were placed in mixed-institution teams designed to promote cross-cultural interaction and shared problem-solving.The team structure also offered exposure to diverse perspectives.Source: A. Participants · p. 3

B. Instructional Design of the VGC Activity

The VGC activity used a structured six-phase project that joined cultural orientation with shared data work and reflection. Teams coordinated asynchronously using online communication and computational tools.

B. Instructional Design of the VGC Activity

  • The project was deliberately structured around clear tasks, role clarity, and scaffolded collaboration across cultural and institutional boundaries.Source: B. Instructional Design of the VGC… · p. 3
  • Students began with cultural data points and contextual videos, then selected an open-repository dataset and set up a shared Google Colab workspace.The orientation invited reflection on how culture influences communication and teamwork.Source: B. Instructional Design of the VGC… · p. 3
  • Teams cleaned and documented data, developed analysis ideas, and divided visualization and machine-learning tasks across institutions before comparing and interpreting results.Source: B. Instructional Design of the VGC… · p. 3
  • Teams reported their work using a Question–Analysis–Conclusion structure, completed reflections, and submitted final project artifacts.Students also shared cultural reflections on a visual board.Source: B. Instructional Design of the VGC… · p. 3
  • Microsoft Teams supported communication and Google Colab supported coding and analysis; interaction was asynchronous, with optional synchronous meetings for coordination.Source: B. Instructional Design of the VGC… · p. 3

C. Survey Instrument

The survey assessed collaboration processes and learning outcomes through construct-organized Likert items. Responses used a five-point scale, with the full instrument provided in Appendix VII.

C. Survey Instrument

  • The survey organized Likert-scale items into seven constructs derived from the conceptual framework.The listed constructs include Collaborative Experience, Communication and Cultural Dynamics, Logistical Strategies, Learning and Skill Development, Global Citizenship and Readiness, and Challenges and Opportunities.Source: C. Survey Instrument · p. 3
  • Responses ranged from Strongly Disagree (1) to Strongly Agree (5), capturing collaboration processes and learning outcomes.The complete survey instrument, including Likert-scale and open-ended items organized by construct, appears in Appendix VII.Source: C. Survey Instrument · p. 4

D. Data Collection

Survey responses were collected electronically, then analyzed with non-parametric tests, regression, and inductive coding. All seven survey constructs met the stated reliability threshold.

D. Data Collection

  • Survey responses were collected electronically after the activity; participation was voluntary and anonymized, with informed consent obtained.Source: D. Data Collection · p. 4

E. Data Analysis

  • Non-normal construct distributions led the authors to use non-parametric tests, including Wilcoxon, Mann–Whitney U, and Spearman tests.They also used standardized multiple regression models, checked multicollinearity and residual normality, and compared U.S. (n = 21) and Pakistan (n = 15) cohorts.Source: E. Data Analysis · p. 4
  • Open-ended responses were inductively coded against the seven constructs, with two additional researchers reviewing codes and resolving discrepancies through discussion.Theme frequencies were reported alongside quantitative results.Source: E. Data Analysis · p. 4

A. Reliability Analysis

  • All seven constructs exceeded the reliability threshold of α ≥.70, ranging from .757 (CC) to .931 (GCR).CC1 had the weakest item-total correlation (r = .361) and would marginally improve the CC scale if removed, but was retained for content relevance.Source: A. Reliability Analysis · p. 4, Table I · p. 4

B. Descriptive Statistics

Students gave positive ratings across constructs, with no significant differences between the U.S. and Pakistan cohorts. Tests against the neutral midpoint supported the three mediator and four outcome hypotheses.

B. Descriptive Statistics

  • All construct means exceeded 4.0 on the five-point scale, above the neutral midpoint of 3.0; PV was highest (M = 4.478) and CC lowest (M = 4.067).All seven constructs showed non-normal distributions (p < .05), consistent with a ceiling effect; the authors therefore used non-parametric tests.Source: B. Descriptive Statistics · p. 4, Table II · p. 4

C. Group Comparison: U.S. vs. Pakistan

  • No construct differed significantly between U.S. (n = 21) and Pakistan (n = 15) students (all p > .05, all effect sizes small), although Pakistan students rated constructs slightly higher.Pakistan students showed lower variability, including SDP V = 0.183 versus 0.628 for U.S. students; the authors concluded positive findings held across both institutional contexts.Source: C. Group Comparison: U.S. vs. Pakistan · p. 5, Table III · p. 5

D. Hypothesis Testing: H1–H3

  • All four outcome constructs exceeded neutral significantly (all p < .001, r = .835–.872), and the three mediator constructs were also significantly above neutral.The mediator tests supported H1–H3; the reported effects were large (r > .82, all p < .001).Source: D. Hypothesis Testing: H1–H3 · p. 5, Table IV · p. 5

E. Hypothesis Testing: H7

Students rated personal growth and opportunity above project challenges. Correlations and regression linked collaboration processes with outcomes, while qualitative themes included technical learning, growth through struggle, and global-work confidence.

E. Hypothesis Testing: H7

  • Students rated personal growth and opportunity items higher (CO2–CO5; M = 4.201) than challenges (CO1; M = 3.917), a significant difference (W = 301.5, Z = 2.243, p = .012, r = .374).The item describing the experience as overall more rewarding than challenging had the highest CO mean (M = 4.389, Mdn = 5.000); H7 was supported.Source: E. Hypothesis Testing: H7 · p. 5

F. Correlational and Regression Analysis: H4–H6

  • All 21 construct-mean correlations were significant (p ≤.006); all 12 mediator–outcome pairs were positive, with 11 showing large effects (ρ = .572–.760).CC and LS were highly intercorrelated (ρ = .887), as were PV and GCR (ρ = .883); CE–CO was the only medium-effect mediator–outcome pair (ρ = .445, p = .006).Source: F. Correlational and Regression… · p. 5
  • The mediators jointly explained 54.8% of variance in LSD, 44.7% in PV, and 52.1% in GCR; CC was the only significant GCR predictor (β = .669, p = .013).No individual predictor reached significance in the LSD or PV models, likely reflecting high CC–LS intercorrelation (ρ = .887) and the modest sample (n = 36); the LSD model also had a residual-normality violation (W = 0.897, p = .003).Source: F. Correlational and Regression… · p. 5, Table V · p. 5

G. Qualitative Themes

  • Technical Skills Gained was the most frequent qualitative code (n = 20), while Growth Through Struggle dominated Challenges and Opportunities (n = 19).Positive Experience was the leading CE theme (n = 22); Diversity as Asset (n = 15) and Confidence in Global Work (n = 12) led GCR, while time-zone difficulty (n = 11) usually appeared alongside growth-oriented codes.Source: G. Qualitative Themes · pp. 5–6

VI. DISCUSSION

Across both cohorts, students rated the collaboration positively; collaborative processes were linked to learning and perceived value, while cross-cultural communication uniquely predicted global readiness.

VI. DISCUSSION

  • All seven hypotheses were supported, with findings consistent across the U.S. and Pakistan cohorts.Source: §VI · p. 6

A. Students Rated the Experience Positively

  • Students rated collaboration, communication, and logistics well above neutral, with similarly positive ratings across cohorts.The discussion reports large effect sizes for these ratings.Source: A. Students Rated the Experience… · p. 6
  • Pakistan students reported slightly higher scores with less variability; positive-experience codes dominated, while unequal-participation codes were few.The discussion suggests partner disengagement was isolated rather than systemic.Source: A. Students Rated the Experience… · p. 6

B. Learning and Value Were Driven Collectively

  • Collaboration mediators together explained roughly half to nearly 60% of learning and perceived-value variance, but CC and LS predictors could not be separated individually.Students often perceived communication and logistical coordination as closely connected.Source: B. Learning and Value Were Driven… · p. 6
  • Dominant themes included coordination, task management, communication practices, technical skills, and cross-cultural collaboration, spanning process and learning gains.The reported learning gains included both technical and interpersonal dimensions.Source: B. Learning and Value Were Driven… · p. 6

C. Communication Drives Global Readiness

  • Cross-cultural communication uniquely predicted global citizenship and readiness (β = .669, p = .013); students also described cultural diversity as an asset and technical work as a bridge.These qualitative themes were presented as explaining how communication across cultures relates to readiness.Source: C. Communication Drives Global Readiness · p. 6

D. Challenges Were Seen as Worth It

Students acknowledged real collaboration challenges but generally described growth and opportunities as outweighing them; the paper also reports study limitations and directions for future work.

D. Challenges Were Seen as Worth It

  • Students across both cohorts described growth through struggle; time-zone challenges and partner engagement were the main named difficulties, yet workarounds helped teams complete the project.Growth Through Struggle was the highest-frequency challenge/opportunity code (n = 19); Time Zone Challenges had 11 mentions and Partner Engagement Issues had 7.Source: D. Challenges Were Seen as Worth It · p. 6
  • Challenges were acknowledged, but students generally rated the collaboration as more rewarding than difficult, yielding a medium rather than large advantage for opportunities over challenges.Source: D. Challenges Were Seen as Worth It · p. 7

VII. LIMITATIONS

  • Optional participation, self-reported measures, one short activity, and a sample of 36 constrain generalizability and causal claims; CC–LS multicollinearity also limited individual-predictor interpretation.The study also notes reduced regression power and a minor residual-normality violation in the LSD regression.Source: §VII · p. 7

VIII. CONCLUSION AND FUTURE WORK

  • The study concludes that short-term structured VGC can support technical learning, collaborative skills, and global readiness in undergraduate computing education.The authors call for larger samples, longitudinal follow-up, broader institutional contexts, and formal inter-rater reliability measures in future work.Source: §VIII · p. 7
  • Cross-cultural communication was the unique predictor of global citizenship and readiness, while the three mediators explained roughly 45 to 55% of outcome variance across models.The conclusion describes cultural diversity as an asset and shared technical work as a bridge across cultural differences.Source: §VIII · p. 7

APPENDIX

The appendix identifies the complete survey instrument and specifies the response formats for its item types.

APPENDIX

Loading 2610.01147v1…