Retain or improve a research Method
Actual research work, criticism, a changed tool or new receiving conditions can raise a Method question. First distinguish the reusable Method from its description, implementation, available capability and the particular analysis or claim. A disappointing finding or software release alone does not establish that the Method failed.
| Your question | Open | What you can obtain |
|---|---|---|
| Should we retain, correct or change this research Method, and what follows for existing results? | RMP.9 - Decide Whether and How to Improve Research Methods and Practice | Supported retention, a verified bounded correction, a justified Method change, or a restriction on unsupported use. When qualification remains unresolved, the result may instead select worthwhile obtainable evidence while preserving that unresolved question. |
Bring the research use, difficulty, current Method and available basis. Finish with an adequate existing answer or determinate correction. Select a further comparison only when different results could change the decision, the comparison is attainable and its whole burden is warranted. Declining that work does not establish Method adequacy; keep unsupported use restricted.
For example, an accepted sample-variance Method for n independent observations divides the squared-deviation sum by n−1. If software uses n instead, correct and check the implementation and its dependent analysis. Under those premises, the Method can stay unchanged without another efficacy study. If the receiving situation instead violates the Method’s assumptions, that correction cannot qualify the new use. A warranted comparison can be selected, but its planned result is not evidence already obtained.
Use Method Engineering only for the identity, trial, fit, worth, variant or introduction contribution this question needs. Preserve the research purpose and its appropriate criticism; a common score need not fit statistical, interpretive and historical inquiry.
Return changed operations to RMP.3, affected records to RMP.4, a changed analysis or open credibility choice to RMP.5 or RMP.6, and changed synthesis to RMP.7. Reopen only the contribution actually affected; return an affected claim or model to RMP.8. A correction confined to an analysis or claim stays with RMP.5 or RMP.8. Reopen RMP.1 only when the Method difficulty meets its research-membership conditions. The decision to change a Method, its actual introduction and evidence of its effects remain different results.