Project method: from a problem to a result, with fewer round trips

The review behind this repository, a first pass of 38 sources. It asks which steps written before the work starts are measured to cut rework, and what a finished job must leave for a second person to resume it.

• It holds 40 sources: 35 formal and 5 grey.
• Grey: 2 score 10 or more of 20 after a second rating: Schwaber 2020 at 10 and Chapman 2000 at 14.5.
• 2 grey sources score below it and are examples of practice only: Singer 2019 at 9 and Ubl 2020 at 9.5. 1 is located, not scored. None is a controlled study.
• Formal: 5 read in the text or the sections cited, 13 at abstract level, 1 through secondary summaries, 16 located with the text not opened.
• It is a first pass. Read the limits in protocol before relying on a number.

Files

File	Role
frame	step 1: the problem in one sentence, the field's names beside the repository's, the four questions as sub-questions
protocol	why grey literature, every search string with its count, the stopping rule, selection, scoring, limits
sources	the pool: how each source was found, what was read of it, its grade or score
review	step 2: the findings by strand, existing reviews first
state of the art	step 3: best evidence apart from most adopted practice, per question
against	the evidence against: contested figures, weak measurement, reversals
position	step 4: what is known, what is not found, what this repository can claim
baseline	step 5: the named baseline, B1, and how to compare against it
bibliography	every work cited, once, with a stable link

Claims

Claim	Grade	Where
Written scope and objectives before building cut rework	convention, untested: no study isolates them	review.md Q1
Avoidable rework is a large share of effort: 40 to 50 percent	expert report, not a measurement	review.md Q1; against.md
A defect found late costs 100 times more as a constant	not supported: 171 projects found no delayed-issue effect (secondary)	against.md
Reading the prior studies before a new one avoids waste	convention, untested for effect; the neglect is measured, descriptive (secondary)	review.md Q1
A named comparison target is part of the process models	design science names it at the objectives and evaluation steps; no model has a review step	review.md Q1; position.md
Writing the plan as a topic changes the outcome	not tested; an outline raises writing quality in one experiment (secondary)	review.md Q2
A fixed set of files lets a second person resume	convention, untested; practice guides agree, no study	review.md Q3
Method ends where the genre starts: communication is the last activity	convention, in the process model	review.md Q4

See also

The front page; Evidence.