CQR — thesis stress among graduate students

Raw data sample_cqr_thesis_stress.csv · 36 rows

Entirely synthetic — not real interviews, and unrelated to real people or institutions. The six cases were written from situations common in the literature on thesis stress.

The study, step by step — what was actually produced

1. ① Coding domains and core ideas

   Two coders each coded all 36 turns without seeing the other’s work. Twenty units were coded by both so that agreement could be computed.

   Inter-coder agreement — Cohen’s kappa = 0.78 (20 double-coded units). The 4 disagreements were carried into the next step as items for consensus.

2. ② Consensus

   The 4 disagreements were laid side by side and settled by exchanging grounds — not by a majority vote, but by the exchange itself.

   The representative case — C3: “Office worker by day, graduate student by night — and I often think I am doing neither properly.” Coder A read it as time pressure, Coder B as self-doubt. They agreed that the focus of the account is an appraisal of the self rather than a shortage of time, and settled on self-doubt.

3. ③ Auditor check

   An auditor who took no part in the analysis reviewed the codebook and the assignments. Of two comments, one was accepted and one was declined; both outcomes were recorded.

   Comment 1: “One coping category holds counseling, peer groups and exercise together — what is leaned on differs, so counting them separately sits closer to the data.” → Accepted: split into “leaning on people” and “moving the body”. Comment 2: “Could comparison with peers be folded under self-doubt?” → Declined: comparison is a separate current arising in a relational context, and its case pattern differs (4/6 against 6/6).

4. ④ Cross analysis

   We counted in how many cases each domain appeared — the unit is how many people it came from, not how many times it was said.

   Out of 6 case(s) — General ≥90%, Typical ≥50%, Variant <50%. Result — General 2 · Typical 3 · Variant 2.

5. ⑤ Results

   The frequency table by domain (G/T/V) and the quotation table are laid out together with kappa, the consensus record and the auditor record. The process of reaching consensus is itself what makes this method trustworthy.

Key results

Representative quotes

Every quote is a sentence that exists in the raw data (excerpt-existence check passed).

• “I write from ten at night after work”

  Time pressure · C3

• “When I could not follow a question in a seminar, I found myself wondering whether I am someone who belongs here”

  Self-doubt · C4

• “While I wait I do not know whether the direction is right, so I cannot move ahead”

  Weight of the advising relationship · C1

• “I cried in front of the counselor for the first time, and after that I sleep a little”

  Coping — leaning on people · C5

• “Some nights I wake at dawn thinking about the thesis and meet the morning that way”

  Bodily signals · C1

Paper-ready tables