A Total Is a Compressed Judgment
A grant score records how reviewers applied specified criteria to the evidence. It is not the application itself, a probability of success or automatic authority to award. Understand the scheme's actual criteria, scoring method, eligibility and decision rules before interpreting a ranking.
Some schemes use weighted totals. Others use an overall judgment informed by criteria without mechanically summing them. Never import another institution's scale or assume high numbers always mean better. The fictional points below exist only to teach information loss; they are not a generic NGO scoring system.
The Same Total Can Hide Different Concerns
A fictional call publishes three maximum point allocations: method feasibility 40, relevant community fit 40 and resource realism 20. It also states a separate applicant eligibility requirement. Two eligible applications receive these illustrative panel judgments:
| Application | Method /40 | Community Fit /40 | Resources /20 | Total /100 |
|---|---|---|---|---|
| A | 38 | 25 | 18 | 81 |
| B | 29 | 36 | 16 | 81 |
The sums are 38 + 25 + 18 = 81 and 29 + 36 + 16 = 81. The equal totals do not show equal strengths. A has a more convincing method but weaker evidence of local fit; B has stronger local fit with more method questions. Inspect the reasons and material uncertainty, not just a tie label.
These points are reviewers' judgments under a fictional published scheme, not measurements with laboratory precision. A one-point difference does not automatically establish an important substantive difference. If the scheme prescribes tie-breaking or minimum criterion requirements, use those actual rules rather than invent one after seeing the applicants.
Keep Eligibility and Uncertainty Separate
Suppose application C would score highly on quality but does not meet a published applicant requirement. A favorable quality score does not cure ineligibility. Conversely, an application missing evidence needed to establish eligibility is unresolved, not necessarily proven ineligible. Follow the actual clarification and decision rules consistently.
Likewise, a polished budget can be uncertain if a key site permission is unconfirmed. Record which assumption drives the rating and what evidence is absent. Do not disguise a mandatory safeguard as a small numerical penalty that can be offset by strong writing elsewhere.
Use Discussion to Improve the Reasons
Ask reviewers to point to the application evidence, criterion and explanation behind a rating. Correct factual mistakes, obtain appropriate expertise for questions outside the panel's competence, and preserve defensible differences in judgment. A chair can facilitate a fair discussion without instructing everyone to give identical scores merely to remove disagreement.
If the actual method requires an overall judgment, explain how the criteria inform it rather than reverse-engineering a formula. If it requires a weighted calculation, check the formula and units. A consistent calculation can still rest on inconsistent interpretation.
Try criterion calibration and reasoned reviewer discussion. For selection beyond ranking, inspect the decision-record example: only the actual published process and authorized roles determine what other considerations may be used.
Use the existing evidence rubric and panel record to record the actual call criteria and reasons. This explanation adds why a total can conceal different profiles; it does not replace that working record.
Source and Scope
First-Level Peer Review, updated April 28, 2026, distinguishes criterion scores and overall judgment in US biomedical review. Its scale is not adopted here. The fictional comparison demonstrates compression and uncertainty without prescribing an external scoring standard.