Part I of this series documented what ten years of pistol qualification data reveals about organizational performance. The movement gap is real. The regression problem is real. The spread from 35.40 to 44.00 mean score across twelve entities is real. What the analysis could not do, structurally, not methodologically, is explain any of it. The scores were hand-entered by instructors observing targets in the field. That constraint does not disappear with better analysis. Baked into the record before the analyst ever sees it.
What a Score Is, and What It Is Not
A score of 7 on a movement stage tells you the shooter completed the string and earned seven points. It does not tell you where on the target those rounds landed. It does not tell you whether the group opened horizontally, a grip problem, or vertically, which points to trigger control degrading under movement stress. It does not record split times, sequence, or which hits came from which position in the string.
What manual entry produces is a number. What training management needs is a record. The score is the shadow of the event, not the event itself. In the absence of the event, the shadow is the only thing that remains.
A score of 7 on a movement stage tells you the shooter passed. A target image tells you whether they’ll pass next time. And why they might not.
The Regression Problem, Revisited
Part I found that 40.5% of repeat shooters declined in score across their qualification history. Part II adds the harder finding: the decline is unknowable from the record alone. Was the degradation concentrated in movement stages or distributed evenly? Did it accumulate gradually or appear between two sessions? Did the group pattern shift, suggesting a technique change, or simply widen, suggesting atrophy?
Manual scoring cannot answer any of these questions. The answers existed on the targets that were scored and discarded after each session. A training director looking at a shooter who dropped 18 points across three appearances has a score trend and nothing else. They cannot prescribe a fix. They can only observe the decline and hope the next qualification looks different. Most of the time, without a diagnosis, it does not.
The Pencil-Whip Problem Is a Liability Problem
Manual scoring places the entire burden of record integrity on a single person with a pen. The outcome is structurally predictable. Instructors know their shooters. They train them, work alongside them, and are responsible for their deployment readiness. When a shooter posts a 34 in a system where 35 is the floor, the social architecture of the qualification environment produces a specific pressure. Rounding that score is not a character failure. It is what the system was designed, inadvertently, to make easy.
A target-based record removes the decision entirely. The target records what happened. The score reflects what the target shows. The instructor’s judgment is no longer the chain of custody.
As agencies begin adopting verified, image-backed qualification records, the standard of care for documentation shifts across the field. An agency that continues to rely on hand-entered scores while peers maintain reviewable, timestamped records is making a documented choice. One that becomes relevant in use-of-force litigation, negligent training claims, and regulatory audit. The question is not whether better record infrastructure produces better data. It does. The question is when the absence of that infrastructure stops being an administrative gap and starts being an institutional liability.
The target records what happened. The score reflects what the target shows. No one has to decide whether to round.
What the record captures — manual entry versus target-based documentation
| Data point | Manual entry record | Target-based record |
|---|---|---|
| Score per stage | Instructor-assigned integer | Computed from verified hit locations |
| Hit location | Not recorded | X/Y coordinates per round |
| Group pattern | Not captured | Horizontal vs. vertical dispersion |
| Score auditability | Instructor word only | Target image on file, reviewable |
| Pencil-whip risk | Undetectable without witness | Eliminated — score derived from evidence |
| Failure diagnosis | Score below threshold | Where, how, and in which condition |
| Longitudinal trend | Score drift only | Pattern migration across cycles |
What the Dataset Could Have Told Us
The dataset in Part I is a serious institutional record. The findings are real. But every one of them is a finding about scores, not performance. Those are related but distinct. A score is a number someone wrote down. Performance is what happened on the target.
A parallel dataset built on target-based records would have answered different questions. Not just whether the movement stage average was 7.57, but where those points were being lost. Not just that 40% of repeat shooters declined, but whether the decline was concentrated in a specific stage, a specific hit zone, a specific condition. Not just that Group HH had a 60% pass rate, but whether that reflects a training problem or a scoring consistency problem. Two diagnoses with entirely different remediation paths.
The dataset we have is a record of what qualified and what did not. The dataset this methodology makes possible is a record of what happened and why. For organizations that treat qualification as a genuine competency benchmark rather than an administrative requirement, that distinction is the difference between managing a number and managing a skill.
Every qualification in this dataset produced a target. Only the number survived.
If your qualification records were audited tomorrow, not the pass/fail outcomes, but the evidentiary basis for those outcomes, what would the audit find? A number someone wrote, or a record that shows what actually happened on the target?
Part III of this series examines what that record looks like in practice. What it takes to build qualification infrastructure that survives scrutiny at every level of the training enterprise.