Unit 12.00: A structure that serves two readers at once

Seven sections. The two that get dropped are the two that make it an analysis rather than an argument.

A structure that serves two readers at once

Question states the decision in one sentence. Data gives the source, the window, the unit and the sample size before and after exclusions. Method names the test and what was pre-specified. Result gives the estimate with its interval in real units. Uncertainty says what the study could and could not have detected. Limitations names the confounders, the missingness and what was not tested. Next step says what the evidence supports.

The structure serves two readers who never read the same parts. Someone deciding reads Question and Result; someone checking reads Method and Limitations. Both paths have to work alone.

Uncertainty and Limitations are the ones that vanish under deadline pressure, and they are what distinguishes a report from advocacy.

This block lays out the seven sections and what each must contain.

sections <- c(
  "Question"    = "One sentence. What decision does this inform?",
  "Data"        = "Source, time window, unit of analysis, n before and after exclusions",
  "Method"      = "The test, why it was chosen, what was pre-specified",
  "Result"      = "Estimate with its confidence interval, in real units",
  "Uncertainty" = "Interval width, sample size, what the study could have detected",
  "Limitations" = "Confounders, missingness, generalisability, what was NOT tested",
  "Next step"   = "The decision this supports, or the evidence still needed"
)
for (nm in names(sections)) cat(sprintf("%-12s %s\n", nm, sections[[nm]]))

cat("\nA reader in a hurry reads Question and Result. A reviewer reads Method\n")
cat("and Limitations. Both must work without the other.\n\n")
cat("The two sections most often missing are the last two -- and they are the\n")
cat("ones that separate an analysis from an argument.\n")
cat("Sections:", length(sections), "  None is optional.\n")

7 sections, none optional. The output makes the reader split explicit — Question and Result for the hurried, Method and Limitations for the reviewer — and names the two most often missing as the last two.

The mistake this prevents

The mistake is folding limitations into a closing sentence of the result. They belong under their own heading, where a reviewer looks for them and cannot skim past.

Takeaway

Use all seven headings. Write Limitations before you write the conclusion, and make sure Result and Method each stand alone.

Mark complete