teaching-phd-skills

SEAR Lab Research Handbook

Start here.

This is the handbook for research skills in the SEAR Lab at UT Arlington. If you are a new PhD student, everything you are expected to be able to demonstrate is on this page, with a lesson behind each one.

How this works

You keep a weekly research log in three channels — reading, writing, arithmetic. That is graded every week and it is not going anywhere.

A skill below is done as one week’s log entry, not in addition to it. Pick one, do it on your own dissertation work, submit it in the channel it belongs to.

A milestone is a group of related skills. When they are all in, you submit an index listing which week each went in. It is graded complete / redo. Once you have passed a skill you never repeat it — not this semester, not in year four.

Target two milestones a semester. Twenty-eight skills across six milestones takes three to four semesters, which puts you finished around the time you are proposing.

What that looks like in practice

The part that is easiest to get wrong is thinking a skill is something you hand in separately. It is not. Here is a whole milestone, start to finish.

Week 7. You pick Git and GitHub from M2. You read the lesson, then spend an afternoon doing it on your actual thesis repo — writing the .gitignore, verifying it, making the first commit.

Week 7, Tuesday night. You open WA7 in Canvas — the arithmetic entry you were going to submit anyway — and instead of writing about your week, you write about this. In the text box: what you nearly committed and did not, plus your repo URL and the pasted git check-ignore -v output. Submit. WA7 is now done. You did not submit anything extra, and there is nothing else to hand in this week.

Weeks 9, 11 and 13. Same thing for the other three M2 skills — code folder structure, reproducible environments, CLI agents — each one becoming that week’s WA entry. They have to be different weeks; you cannot do all four in one.

Week 13, after the fourth. Now you open the M2 — Reproducible Code assignment. It is an index, not new work. Four lines:

Git and GitHub            — WA7,  week of 6 Oct   — github.com/you/thesis
Code folder structure     — WA9,  week of 20 Oct  — same repo
Reproducible environments — WA11, week of 3 Nov   — run log attached
CLI agents                — WA13, week of 17 Nov  — transcript in that entry

Plus one paragraph on which was hardest. Submit.

Then I grade it complete or redo. Redo means one of the competency checks did not pass — I will say which, and you redo only that one skill, not the milestone. Once M2 is complete it is complete permanently.

Every lesson ends with a Turning it in section that spells this out for that specific skill: which channel, what goes in the text box, what to attach.

Two rules

Everything is done on your own work. Your actual thesis repo, your actual draft, your actual data. Not a sample.
Where the lab code standard governs something, the lesson cites the Part and stops. Go read the Part. This is deliberate: a summary reads as complete and stops you looking, so you will not find one here.

Nothing is gated. The tiers below are advice about a sensible order, not locks. If your research needs M5 in year one, take M5 in year one.

Choosing what to do this week? the whole curriculum on one page puts all twenty-eight on one scrollable page — what each is for, what you hand in, and the checklist it is graded against — so you can compare them without opening twenty-eight files. It is generated from the lessons, so it cannot go stale.

M1 — The Workbench · Year 1

The five surfaces you actually work on, plus the habit that makes all of them safe: checking a result against something you knew independently.

Skill	Channel	Turn in as
Spreadsheet discipline	arithmetic	`.xlsx` plus annotation
Notebooks and AI assistance	arithmetic	notebook link plus verification table
The command line and a remote machine	arithmetic	terminal transcript plus retrieved output
LaTeX and Overleaf	writing	Overleaf link plus compiled PDF
Sanity checks and orders of magnitude	arithmetic	text entry

Take the command line before M2 — skills 06 and 09 both assume it.

M2 — Reproducible Code · Year 1

The milestone that prevents “I cannot reproduce my own Chapter 3.” The test at the end is not a file listing — it is that another student ran your code on their machine and got your number.

Do Git and GitHub first. Git history is permanent, and a credential committed once is compromised no matter what you do next. It is the only thing in these twenty-eight skills that cannot be undone afterwards.

Skill	Channel	Turn in as
Git and GitHub	arithmetic	repo URL + `git check-ignore -v` output
Code folder structure	arithmetic	repo URL + what-moved paragraph
Reproducible environments	arithmetic	repo URL + partner’s signed run log
CLI agents and Claude Code	arithmetic	transcript excerpt + accepted diff

M3 — Reading and Reasoning · Year 1–2

Evaluating somebody else’s argument, finding where the field stops, spotting an inference that fails, building a claim of your own, and saying it no more strongly than your evidence allows.

A claim needs reasons and evidence. The link between them is a warrant. Warrants are almost always unstated — and the unstated warrant is where an argument actually breaks.

On “critical thinking”

You will be asked whether your training covered this — in interviews, on fellowship applications, by your own committee. The answer is yes, and the useful version of the answer is specific, because the phrase on its own means almost nothing and everybody claims it.

This milestone is where it lives, and each skill teaches a different part:

Analysis — taking an argument apart: what is claimed, what is assumed, what is actually shown (critiquing papers)
Evaluation — judging whether an inference holds, including when it is fluent and wrong (logic and its failures)
Synthesis — putting fifteen papers on one axis to see where the field stops (gap-finding)
Creative thinking — generating framings and choosing between them, rather than defaulting to the first one you saw (gap-finding)
Constructing an argument — a claim with reasons, evidence, and a warrant you can state (claim, reason, evidence · saying exactly what you claim)

Say that instead of the phrase. It is more impressive, and it is true.

Skill	Channel	Turn in as
Reading and critiquing papers	reading	WR entry, 300–500 words
Literature review and gap-finding	reading	three candidate axes, the map, the gap
Logic and its failures	reading	three worked examples
Claim, reason, evidence	writing	one page
Saying exactly what you claim	writing	claim-verb audit of a draft and your CV

M4 — The Manuscript · Year 2

Everything that goes into assembling a document somebody else can read, trust and cite. The argument itself was M3; this is the artifact that carries it. All five are five passes over the same draft.

Skill	Channel	Turn in as
IMRAD structure	writing	Overleaf link
Citation verification	reading	table plus updated draft
Research ethics and authorship	writing	CRediT statement plus evidence circulated
Figure and table design	writing	PDF: before, after, caption
Accessibility	writing	repo URL

M5 — The Review Cycle · Year 3+

Everything between “the draft is finished” and “it is out in the world.” You referee a labmate’s Report Draft 1 and they referee yours; each of you then answers the review you received.

Skill	Channel	Turn in as
Reviewing papers	reading	structured referee report
Response to reviewers	writing	PDF response letter
Venue and audience selection	writing	memo plus cover letter or outreach message
Conference talk and poster	writing	slides plus three questions
Defending your work	writing	three steelmanned objections, delivered live

Defending your work does not wait for year 3. If you are proposing next semester, take it now — it is the one skill with a hard external deadline attached.

M6 — Stewardship · Year 3+

Four things that outlast you here: a citable deposit, a dataset somebody can still find the terms for in four years, an idea written well enough to be funded, and a person who learned to do research because you scoped their first task properly.

Skill	Channel	Turn in as
Zenodo deposit	arithmetic	record URL
Research data management	arithmetic	one-page data statement
Proposal writing fragments	writing	PDF to the call’s page limits
Mentoring an undergraduate	writing	task brief plus retrospective

Two pages that are not graded

The twenty-eight skills above are general — true of a PhD in almost any field. These two narrow from there, and neither is assessed. They differ in when you should read them.

SEAR Lab Domain Knowledge — read this in your first week. The lab’s organising thesis that infrastructure has three layers, the three research areas laid against them, and which course goes deep on each. Surface level on purpose; the courses are where you learn it. Almost nothing else here makes sense until you have this.

Engineering Research Domain Knowledge — read this later, once the domain has stopped feeling new. Philosophy of science, the normative content of modelling choices, policy literacy, the history of your field, the economics underneath it. A discount rate is a claim about how much future people matter; this is the page about that sort of thing.

And one bonus lesson

The model hierarchy: pencil, spreadsheet, simulation is optional and belongs to no milestone. Solve the smallest version of your problem by hand, reproduce it in a spreadsheet, then switch your full model down until it has to agree. It is how this lab checks a model, including one an AI wrote for you. Best done before you build anything large; if you do it, it is one week’s WA entry.

If you are brand new

Read SEAR Lab Domain Knowledge. Half an hour, and everything else lands better.
Read the code standard’s GETTING-STARTED.md.
Do Git and GitHub before you commit anything, anywhere. The ordering in that lesson is the point.
Then pick anything from M1 or M2.

Maintenance

Senior students maintain this. If a lesson is wrong, out of date, or assumes something you did not have, open a PR — that is the intended path, and the contributing notes say what an exemplar needs before it can go in.

This site is open source. Improve this page.