Condition records labeled “intelligence” in Claim Raven's Board dataset have a 42.7% favorable-outcome share. Records labeled “logistics” have a 32.5% share. The apparent gap is 10.2 percentage points.
But the intelligence group contains just 131 condition records from 73 distinct decisions. The logistics group contains 3,286 records from 1,875 decisions. And most records in the wider database have no military-job category at all.
Those details change how I read the comparison. It describes labels and outcomes in a selected sample. It does not show that an intelligence veteran has better odds than a logistics veteran, or that one occupation produces stronger claims.
The seven-category comparison
The table covers 18,905 condition records from 10,140 distinct Board decisions, checked on September 15, 2026. A condition record is one condition analyzed within a decision. A single decision can contribute several records.
| Extracted job category | Condition records | Distinct decisions within category | Favorable outcomes | Favorable share |
|---|---|---|---|---|
| Intelligence | 131 | 73 | 56 | 42.7% |
| Administration | 1,709 | 1,052 | 631 | 36.9% |
| Technical | 5,443 | 3,188 | 1,986 | 36.5% |
| Combat arms | 5,738 | 3,213 | 2,078 | 36.2% |
| Aviation | 1,540 | 953 | 555 | 36.0% |
| Medical | 1,058 | 569 | 375 | 35.4% |
| Logistics | 3,286 | 1,875 | 1,068 | 32.5% |
“Favorable” follows Claim Raven's shared definition: granted, partial-grant and mixed outcomes. The denominator includes those outcomes plus denials and remands. A remand is included as a separate outcome, not as a grant. Dismissed, withdrawn, moot and unclassified outcomes are excluded from the calculation.
The rows are broad categories extracted from the decision text. They are not a verified roster of official Military Occupational Specialty codes. MOS is an Army and Marine Corps term; other services use their own occupation designations. Our categories are a simplified cross-service grouping.
For the larger dataset and definitions, see our VA disability statistics reference.
Why these are not job-based approval odds
The full database contains 101,518 condition records from the analyzed Board-decision subset. Of those records, 82,534 have no extracted job category. Only 18,905 fall into the seven categories shown here; another 79 use other labels, including “unknown.” We have not shown that the classified records represent the missing ones.
The decisions are not independent observations of separate people, either. A veteran appealing several conditions can contribute several records. The decision counts in the table also overlap: 727 of the 10,140 decisions have condition records assigned to more than one of the seven categories. That may reflect multiple service roles, extraction inconsistencies or both. This analysis has not resolved each instance.
Finally, these are matters that reached the Board and were captured in our dataset. They do not describe every initial VA claim filed by people in each occupation.
Calling the logistics percentage reliable merely because that group is larger would miss those problems. More rows do not correct a selected sample, missing labels or uncertain classification.
The table does not explain the gap
Different condition mixes, service periods, types of appeal and evidence may contribute to the observed differences. This table has not isolated their effects.
It therefore cannot establish that intelligence personnel have better documentation, that administrative work produces stronger mental-health claims, or that logistics veterans pay a particular penalty for less complete records. Those are hypotheses that would need a separate analysis and checks against the source decisions.
Even the 10.2-point spread should be read in that light. It is an arithmetic difference between two extracted groups. It is not an estimate of what would happen to the same claim if the veteran had held a different military job.
Your actual duties can still matter as evidence
The limits of an occupation ranking do not make service history irrelevant. 38 CFR § 3.303(a) directs VA to consider the places, types and circumstances of service, along with service records and pertinent medical and lay evidence.
That is where a job history can become useful. A duty title may help identify relevant records, but the details explain what the work involved: the assignment, location, dates, tasks, equipment and any documented event or exposure connected to the claimed condition.
For example, two veterans with a logistics designation may have had very different assignments. One may have worked primarily in an office; another may have regularly loaded equipment or worked around running vehicles. Those are illustrative possibilities, not findings about everyone in that occupation.
The claim needs the veteran's actual history. A broad category should not be used to assume an exposure, rule one out or supply a medical connection that the evidence has not established.
Turn the job history into specific questions
For a condition you are preparing to claim, consider:
- Which assignment and time period are relevant?
- What did the work require you to do, and how often?
- Which personnel, service treatment or other records document those duties or an event?
- Who directly observed something relevant and could describe it accurately?
- What medical evidence addresses the diagnosis and any connection that still needs to be established?
These questions help connect a service history to the evidence needed for a particular claim. They do not presume that every person with the same MOS developed the same condition.
VA's evidence guidance explains the broader requirements. If a statement from someone who served with you would address a specific fact, our buddy-letter examples may help you organize that account.
I would use an occupation category to start locating relevant evidence. I would not let a category's percentage decide whether an individual claim is worth examining.
Sources and limits
This is a descriptive analysis of AI-extracted fields in Claim Raven's case_analysis dataset. The database counts were rechecked on September 15, 2026. The query groups existing labels and applies the shared outcome taxonomy; it does not independently validate every occupation label or outcome against the full decision.
The unit is a condition within a decision, not a unique veteran. Category counts overlap across decisions, missing labels are common, and the sample consists of selected Board appeals. No causal effect, population-wide occupation ranking, confidence interval or individual approval probability is claimed. Further work would need to resolve the category inconsistencies, validate a source sample and account for differences between the groups before making stronger comparisons.
I'm Landon, founder of Claim Raven and an Army veteran. Claim Raven sells claim-preparation tools. This analysis is educational and does not replace legal or medical advice about an individual claim.
When the Board Quotes a Lay Statement: What Our Sample Shows
How to Write a Military to Civilian Resume (With Real MOS Translations)
VA Claim Denied for No Current Diagnosis: Check the Record Before Adding More