04 November 2026

General Check-In

Thinking about the last fortnight:

  • How did preparing the formative work compared with your expectations?
  • What are you finding interesting or challenging in the modules?
  • How did you use discussions with other students to prepare for the formative assessment, if at all?

Today

  • Discuss sources, referencing, and your own contribution.
  • Explore AI strategies for coding, statistics, and writing.

Academic Integrity

Your reader or marker should be able to tell:

  • which ideas, words, data, and code came from elsewhere;
  • what you contributed, changed, and understood;
  • what you actually did and what the evidence supports.

Acknowledging help matters, but acknowledgement does not make a prohibited use acceptable.

Can We Trust These References?

We will inspect six examples together. Some may be fine.

  • Take a moment to inspect each example.
  • Decide whether you would submit each extract as it stands.
  • If not, suggest a repair.

The aim is a traceable link: claim -> citation -> reference -> evidence.

Reference Check 1

More regular sleep timing was associated with better attention scores (Ali & Chen, 2024). Related work has examined study routines and attention using diaries, surveys, and laboratory tasks (Bennett, 2022; Chen & Li, 2024; Hughes, 2021; Morgan, 2023; Patel, 2020).

References

Bennett, R. (2022). Sleep diaries in student research. Journal of Everyday Cognition, 6(1), 10-22.

Chen, L., & Li, M. (2024). Measuring attention during independent study. Journal of Student Research Methods, 11(2), 75-89.

Hughes, A. (2021). Laboratory tasks and sustained attention. Journal of Cognitive Assessment, 4(3), 110-123.

Morgan, R. (2023). Planning study tasks in postgraduate education. Journal of Learning Practices, 8(1), 15-24.

Patel, N. (2020). Surveying students’ study routines. Journal of Learning Research, 7(2), 30-42.

Reference Check 2

Nearly two-thirds of postgraduate students sleep fewer than seven hours a night. Ali and Chen (2024) reported an association between sleep timing and attention scores. In this report, I will examine whether sleep timing is associated with attention scores in my dataset.

Reference Check 3

Tadros et al. (2024) reviewed psychological treatments to improve sleep quality in university students.

Reference

Tadros, M., Newby, J. M., Li, S., & Werner-Seidler, A. (2024). A systematic review and meta-analysis of psychological treatments to improve sleep quality in university students. PLOS ONE, 20(2), Article e0317125. https://doi.org/10.1371/journal.pone.0317125

Publisher’s online version

Screenshot of the PLOS article page showing its title, authors, and publication information.

Source: PLOS ONE article page (CC BY 4.0).

Reference Check 4

Higher smartphone addiction leads to poorer sleep quality in college students (Zhu et al., 2024).

Reference list entry

Zhu, W., Liu, J., Lou, H., Mu, F., & Li, B. (2024). Influence of smartphone addiction on sleep quality of college students: The regulatory effect of physical exercise behavior. PLOS ONE, 19(7), Article e0307162. https://doi.org/10.1371/journal.pone.0307162

Study details: In a cross-sectional survey, 4,670 students completed questionnaires about smartphone addiction, sleep quality, and physical activity. Higher smartphone-addiction scores were associated with poorer sleep quality.

Reference Check 5

Ali and Chen (2024) reported an association between more regular sleep timing and higher attention scores in their sample of postgraduate students.

Reference list entry

Ali, S., & Chen, L. (2024). Sleep timing and attention in postgraduate students. Journal of Student Behaviour, 12(2), 40-51.

Abstract extract: At one visit, 120 postgraduate students reported their sleep timing and completed an attention task. More regular timing was associated with higher scores. Sleep timing was not manipulated.

Reference Check 6

A student asks AI for references about smartphones and attention. It suggests:

Harper, J., & Reed, M. (2022). The distraction advantage: Smartphone presence improves working memory. International Journal of Digital Attention Science, 9(3), 211-225. https://doi.org/10.1086/691462

Would you add this to your reference list? What would you check?

How To Check An AI-Suggested Reference

  1. Find it: search the exact title in Library OneSearch, Google Scholar, or Crossref.
  2. Open it: follow the DOI or publisher link.
  3. Match it: compare authors, title, year, journal, and publication details.
  4. Read it: check the relevant passage supports your claim before citing.

A working DOI is not enough if it leads to a different paper.

Not found does not automatically mean fabricated. Check for errors; do not cite an unverified reference.

Frequent Referencing Mistakes

Pattern Repair
Missing reference Match citations to references in both directions.
Unsupported factual claim Find evidence for the claim, or qualify/remove it.
Incorrect imported details Check authors, year, title, source type, and publication details.
Association presented as causation Match the claim to the design: an association does not establish causation.
Fabricated AI reference Verify existence and metadata.
Unclear citation scope Place citations so readers can tell which claims each source supports.
Citing an unread source Read the original, or explicitly cite it through the source you actually consulted.

What Can Your Referencing Workflow Catch?

How do you currently insert citations and create your reference list?

  • RMarkdown: citation keys link to a bibliography file; rendering produces citations and references in the configured style.
  • Word with a reference manager: Zotero, Mendeley Cite, or EndNote’s Cite While You Write can link citations to an automatically updated reference list.
  • Manual references: check the text and reference list against each other.

Which issues in our examples would your workflow prevent? Which would still need your attention?

Automation helps keep references consistent. It cannot tell whether you forgot a citation, selected the right source, or represented its findings accurately.

From References To AI Use

AI can still misrepresent its findings, population, or limitations.

  • Cite the evidence: read and cite the original sources you actually checked.
  • Acknowledge the process: explain how you used AI, following NTU and assessment requirements. This does not replace evidence for your claims.
  • Verify independently: check the publisher, Library catalogue, or a bibliographic database.

Use AI to help find or understand sources. Cite the sources you checked, and separately acknowledge how you used AI.

What is the equivalent check when AI produces code or a statistical analysis?

Bodleian: verifying references; APA: documenting AI use. Follow your assessment’s specific instructions.

NTU’s Official Position On AI

NTU’s Academic Integrity Policy expects you to:

  • Keep your own contribution clear; do not present AI-generated work as your own.
  • Use AI to support, not replace, your independent work.
  • Declare AI use in assessments.
  • For proofreading, ask for errors to be identified; make corrections yourself.

Disclosure of AI use is not, by itself, evidence of misconduct. Transparency matters; you must still follow the rules for the assessment.

NTU Academic Integrity Policy

Before Using AI For Assessment

Check the guidance for the assessment:

  1. What uses are allowed, required, or restricted?
  2. What skill or contribution is being assessed?
  3. What acknowledgement or declaration is required?

If it is unclear, ask the module team before using AI.

A Real Example: Cyberspace

PSYC40505 explicitly requires a GenAI preamble to the critical essay:

  • Generate approximately 300 words related to the essay topic.
  • Include the output, tool, and prompt.
  • Critically appraise the output in no more than 150 words.

Does it use accurate evidence? Does it analyse, integrate, and question ideas, or just sound convincing?

AI Strategies Across Your Modules

Possible learning uses, subject to the relevant assessment guidance:

Module Ask for help with Check yourself
Visualisation / dashboards Explain a plot or a small Shiny example; suggest a readability improvement. Labels, units, filters, and whether the display is misleading.
Python / machine learning Explain an error; suggest a small test for one function. Expected output, train/test separation, and data leakage.
Computational statistics Explain an assumption; contrast two approaches on a toy example. Suitability for the design, diagnostics, and interpretation.
Cyberspace / project writing Question an argument or suggest search terms. Read the evidence yourself; verify sources and follow task-specific AI rules.

Your Task: Ask Copilot For R Code

Use the public ggplot2::msleep data from our previous tutorial. One row is a species; vore is diet group and sleep_total is sleep in hours.

We want a small table showing, for each diet group:

  • the number of species with usable data;
  • their average total sleep in hours.

Exclude rows missing either diet group or total sleep. Keep the code simple and readable.

Write your own request in plain English before opening Copilot (in Word or similar). You do not need to know which R functions to ask for.

Build A Clear Request

Compare your first prompt with a partner. Does it explain:

  1. Input: which dataset, what one row means, and which columns matter?
  2. Output: what should one result row contain?
  3. Decisions: what should happen to missing values?
  4. Learning needs: short, beginner-readable R/tidyverse code, with each step explained?

Sentence starters: “My data are…”, “I want a table with…”, “Exclude…”, “Please explain…”

Ask Copilot to clarify anything important rather than silently making assumptions.

Generate, Test, And Revise Once

Use Copilot, then inspect the code before running it in RStudio.

  • Does it use the actual vore and sleep_total columns?
  • Is there one result row per diet group, rather than one overall average?
  • Are missing values handled as requested, and are counts sensible?
  • Can you explain one line and check one group’s average independently?

Revise your prompt once to address a missing detail, unclear explanation, or failed check. If it works already, ask for a way to verify it.

Share one change that made the request clearer, and one check you used. What did you still need to decide yourself?

Wrap-Up

  • What did you change in your prompt, and why?
  • What did you check rather than take on trust?
  • What is one referencing or AI-use question we still need to clarify?