2026-40
The letters that held traced the work forward; the letters that fell did not
In the appealed NIW dismissals where the AAO wrote about the recommendation letters, the letters traced the endeavor's downstream impact in 0.7% of decisions and did not trace it at all in 62.6%. In the sustained appeals, they traced it in 83.2%. The record describes two different letters, not two different writers.
## What the record shows
Among the appealed NIW decisions where the AAO characterized the
recommendation letters, the dismissed side and the sustained side read as
two different documents.
On the dismissed side (n=449), the letters did not trace the endeavor's
downstream impact in 62.6% of decisions, traced it partly in 27.2%, and
traced it fully in 0.7% (3 of 449). The letters referenced the specific
endeavor at all in only 17.4%; in 36.8% they did not reference it.
On the sustained side (n=89), the letters traced the impact fully in
83.2% and referenced the endeavor in 92.1%. The AAO credited the letters
in 98.9% of those decisions.
The figures, their query, and the coding contract are registered as
[S95](/methodology), with the treatment and authorship families as S94
and S96.
## Why this is worth a week's attention
The usual reading of a weak letter is that the writer was not senior
enough. This pattern points somewhere else. The dismissed-side letters are
most often discounted for not linking to the endeavor (57.0%) and for
being conclusory (47.7%); independence is the reason in 2.5%. The letters
that held did one specific thing: they named the endeavor and followed it
forward to who would use the work and what would change.
The practical reading is about the brief you give a recommender. Before
asking who should write, decide what the letter has to trace: the
endeavor by name, the work already done inside it, and the path from that
work to a stated downstream effect. A letter that praises the person and
stops there is the shape the record discounts most.
## What it does not say
These are portfolio-level readings of how the AAO characterized the
letters in its own decisions, not blind grades of the letters themselves.
The dismissed side carries a selection effect: the AAO writes about the
letters it contests. Appealed decisions are their own population and not a
sample of filings, and the two sides are two populations, never a rate
between them. It does not predict any outcome or characterize any
individual case.
Receipt: S95 Methodology