technical_se

technical_se(result, data=None, *, source='auto', icc=None, registry=None)

Standard error on each score attributable to assay measurement noise.

Parameters

Name	Type	Description	Default
result		A :class:~falconage.score.FalconResult.	required
data		The :class:~falconage.core.FalconData that was scored. Required for source="probe"; the per-feature spread has to come from the matrix, not from the scores.	None
source	str	"probe", "clock", or "auto" (probe where possible, clock otherwise, and a recorded refusal where neither is available).	'auto'
icc	pd.Series | None	Per-feature ICC, overriding the bundled table. Pass the output of :func:icc_from_replicates to use your own duplicates.	None

Notes

An imputed feature contributes its full between-sample variance rather than the reduced s²(1-ICC). This is deliberate and it is the part of the calculation most likely to look wrong: an imputed value is the cohort mean or a published reference, so it carries no information about the sample in front of you, and treating it as a well-measured probe would make the interval narrower for worse data.