I've spent 15 years on both sides of federally funded research — designing the studies and, in reviewer roles, reading other people's methods sections by the dozen. The proposals that get funded and the ones that don't are often studying equally good ideas. What separates them is almost never the idea. It's whether the methods section gives a reviewer confidence that the team can actually execute it.
Here's what I'm scanning for, in roughly the order I notice it.
1. Does the sample size have a number attached to a decision?
"We will recruit a sufficient sample" is not a power analysis. Reviewers want to see the specific effect size you're powered to detect, where that number came from (prior literature, a pilot study, a minimum clinically or practically meaningful difference — not a rule of thumb), and the assumptions behind it (alpha, desired power, expected attrition, clustering if relevant). If your design is multi-site or has nested structure — students in classrooms, patients in clinics — your power analysis needs to reflect that structure, not treat every unit as independent.
2. Is the analysis plan matched to the actual data structure?
A surprising number of proposals describe a sophisticated design — multi-site RCT, longitudinal follow-up, multiple informants — and then default to a generic "we will use regression" for analysis. Reviewers notice the mismatch immediately. If your data are nested or repeated, say so, and name the modeling framework (multilevel, GEE, latent growth) that will handle it. You don't need to write the full model equation in a proposal, but you do need to signal that you know which one applies and why.
3. Are missing data and attrition addressed before they happen?
Every longitudinal or multi-wave study loses participants. A methods section that acknowledges this upfront — with a plan for how missingness will be assessed (e.g., comparing completers to non-completers) and handled (multiple imputation, full-information maximum likelihood, sensitivity analyses) — reads as a team that has actually run a study before, not just designed one on paper.
4. Does the measurement have evidence behind it?
If you're using an existing instrument, cite the validity and reliability evidence — ideally in the population you're studying, not just the population it was originally validated in. If you're building something new, describe the development and validation plan as its own deliverable, with a timeline. Measurement gets treated as an afterthought more often than any other part of a methods section, and it's usually the first thing a psychometrically literate reviewer checks.
5. Is there a plan for what happens if the primary analysis doesn't work?
Pre-specified sensitivity analyses, alternative model specifications, and a clear statement of what would count as a null or inconclusive result all signal rigor. Reviewers have seen enough studies to know that the primary analysis doesn't always behave as planned — proposals that plan for that are more credible, not less.
The pattern underneath all five
None of this is about jargon or model complexity for its own sake. It's about specificity. Every vague sentence in a methods section is a place where a reviewer has to guess whether you've actually thought it through — and reviewers default to skepticism when they have to guess. The proposals that read as fundable are the ones where every methodological choice is stated, justified, and matched to the actual design.
If you're heading into a submission and want a second set of eyes on the methods section before it goes out, that's exactly the kind of review I do. Get in touch.