7AAVDH10 Chap.6 Research Design and Final Project
Research Design and Final Project
Align the two submitted files
The final project combines an 800-word rationale with one Jupyter Notebook. The rationale frames an original digital humanities question, scholarship, method, findings and limitations. The notebook supplies tangible code evidence through organised cells, comments, validation and visualisations.
This chapter links every objective to a code section and every output to an interpretation.
Demonstrate understanding visibly
The brief values simple code that the student understands over opaque complexity. The chapter maps research and code criteria to visible evidence, records coding-assistant prompts where relevant, and audits the notebook from a clean run before submission.
What this chapter covers
- 01
Research aims
- 02
Research questions
- 03
Academic context
- 04
Notebook structure
- 05
Code criteria
- 06
Submission audit
Worked example · free
Narrow a digital humanities question
- 3Name the cultural object and comparison. The marks shown are a study allocation, not an official University marking scheme.
- 3Choose evidence and a reproducible method.
- 3Connect outputs to interpretation and limits.
Key terms
- Research Aim
- A research aim states the broad purpose the project seeks to advance.
- Research Objective
- A research objective identifies a concrete step used to achieve the aim.
- Reproducibility
- Reproducibility means the documented code and inputs can regenerate the reported outputs.
Research Design and Final Project FAQ
How narrow should the project question be?
It should identify a cultural object, comparison or pattern and the evidence needed to address it within the available project scope. A focused question makes data selection, code choices and interpretation easier to justify.
What shows Python understanding?
Meaningful variables, organised code, comments that explain purpose, small validation tests and the ability to connect loops, conditions and functions to the research method show more understanding than unexplained complexity.
What belongs in the final audit?
Run the notebook from a clean session, verify file loading and saved visualisations, compare outputs with the rationale, remove abandoned cells and match every published criterion to visible evidence in the submitted files.
How do I check alignment between the rationale and notebook?
Take each empirical sentence in the rationale and point to the notebook output, validation and data boundary that support it. Then work forward from every research objective and locate the code section and figure that fulfil it. Empty links reveal unsupported claims, ornamental outputs or objectives that are too broad. Repair the alignment before adding methods.
A clean run should regenerate the cited outputs in the same order and preserve the limitations stated in the writing. Record that final comparison explicitly in the notebook audit, including the expected file names, chart labels and the limitation attached to each interpreted result.
Exam move
Maintain an alignment table linking each research question to an objective, dataset, code section, validation check, output, rationale paragraph and limitation. Begin with a broad interest and narrow it until the cultural object, comparison and evidence are explicit. Test whether the available data and project time can support the proposed scope.
For each objective, name the notebook cells that carry it out and the output that will show progress. Build a criterion map for meaningful variables, sequences, a loop, a conditional, a user-defined function, data loading or export where relevant, cleaning, querying, analysis, at least two visualisations and extensive explanation.
Do not add complexity merely to tick a criterion; choose code that advances the research aim and that you can explain line by line. Keep the visualisations complementary, with complete labels and a written interpretation beside each. Preserve small validation inputs and expected outputs near the functions they test.
If a coding assistant contributes, include the prompt in comments, identify your alterations and explain the purpose and limitation of the resulting block. Keep generated prose out of the rationale under the brief's stated policy. Draft the rationale around question, literature, approach, data, findings and limitations, then point each empirical sentence back to notebook evidence.
Run the notebook from a clean session in order, check paths, remove abandoned cells and confirm saved image files. Compare the final outputs with every statement in the rationale. Finish with an independent submission audit: format, references, objectives, questions, comments, visualisations, code length guidance and file completeness. A criterion is secure only when a reader can locate its evidence without guessing.
Add a weekly reverse-audit that starts with one sentence from the proposed conclusion. Follow it to the supporting visualisation, the table behind that visual, the code that created the table, the validation case and the raw data boundary. Mark any unsupported jump. Then work forward from each research objective and identify the cell and output that fulfil it.
If an objective has no evidence, narrow or remove it before expanding the method. Rehearse explaining one function and one chart without looking at the code, then reopen the notebook and correct the first missing assumption. End by exporting or saving the required outputs to a fresh folder and confirming that another clean run reproduces them in the same order.