Method & Data Sense-making

Act as Reviewer 2 for a Q1 journal in [FIELD].

My study design: [DESCRIBE DESIGN, SAMPLING, INSTRUMENT, ANALYSIS]
My research question: [PASTE]

1. List every threat to validity in this design (internal, external, construct, statistical conclusion).
2. For each threat, say whether it is fatal, fixable before data collection, or acknowledgeable as a limitation.
3. Tell me the three questions a reviewer is most likely to ask that I cannot currently answer.
4. Suggest the minimum change that would most improve rigour without changing my scope.

Turns AI into a pre-reviewer. Finding the fatal flaw before data collection is worth more than any writing speed-up.

I chose [METHOD, e.g. explanatory sequential mixed methods] for a study on [TOPIC] with [SAMPLE].

Write a 200-word methodological justification suitable for a journal Methods section that:
- links the choice to my research question and epistemological stance,
- names the two most credible alternative methods and states why I rejected them,
- uses hedged academic register, no marketing language.

Then list the seminal methodology sources I should cite for this choice (author + year + book/article title only) so I can locate and verify them myself.

The last line is the safe way to get citation leads: names to look up, never a ready-made reference list to paste.

I need a [survey / interview protocol / observation rubric] measuring [CONSTRUCT] among [POPULATION].

1. Break the construct into dimensions, citing the theoretical model I named: [MODEL].
2. Draft 4-6 items per dimension on a [5-point Likert] scale.
3. Flag any item that is double-barrelled, leading, culturally loaded, or too abstract for this population.
4. Suggest 3 items that could serve as attention or consistency checks.
5. Note which items would need reverse coding.

Instrument drafting is generative work on your own construct — no data involved, and the item-flagging step catches classic design errors.

Here is real output from my analysis in [SPSS / R / jamovi]:
[PASTE OUTPUT TABLES]

Sample: [N, design, variables]

1. Explain in plain language what each coefficient, p-value and effect size actually means for my variables — not generic definitions.
2. State the assumptions this test requires and which of my output rows tell me whether they were met.
3. Write the results in APA 7 reporting format.
4. List three interpretations that would OVERSTATE these results, so I can avoid them.
Do not speculate beyond the numbers I gave you.

Point 4 is the quality move: most reviewer complaints are about over-claiming, not about the statistics themselves.

Here is my codebook: [CODE NAMES + DEFINITIONS + EXAMPLES]
Here are [N] anonymised transcript excerpts: [PASTE]

1. Apply my codebook to each excerpt. Quote the exact text segment for every code applied.
2. Where an excerpt fits no existing code, say so and propose a candidate new code with a definition.
3. Where my code definitions overlap or are ambiguous, tell me which pairs are confusable.
4. Do not merge, paraphrase or invent quotes.

I will compare this against my own coding to check consistency.

AI becomes a consistency check on your own coding, not a replacement coder. Report it honestly in your methods.