Documentation
¶
Overview ¶
Package twosample is the one reading of the explicit per-side inputs the Compose two-sample overlays (OVERLAY_T_VS_REF, OVERLAY_Z_VS_REF, OVERLAY_T_CELL, OVERLAY_Z_CELL) take when a side's mean arrives as a bare scalar: its variance and its sample size, from the spec's params.
A side whose mean comes from the slot's Components (an AGG_WELFORD row or a crosstab cell whose components carry mean and variance) needs no params. A scalar side does, and there is NO default: the var = 1, n = 2 fallback once computed p-values from figures nobody supplied (U46 E4-S16). Predict (internal/descriptor) and the runtime handlers (internal/processing) both read params through Inputs, so what one refuses the other refuses.
Leaf: stdlib only. No-execute.
Index ¶
Constants ¶
const ( SymmetricVarianceKey = "variance" SymmetricSampleSizeKey = "sample_size" )
Symmetric keys: one value for both sides, read when the per-side key is absent.
Variables ¶
This section is empty.
Functions ¶
func Inputs ¶
Inputs reads side's (variance, sample size) from params: the per-side key when present, else the symmetric one. A value counts only when it is a finite number > 0; a per-side key that is present but unusable does not fall through to the symmetric key. missing names, in that order, the per-side key of every input that is absent or unusable — nil when both are usable.
func SampleSizeKey ¶
SampleSizeKey is side's sample-size param ("sample_size_target" / "sample_size_ref").
func VarianceKey ¶
VarianceKey is side's variance param ("variance_target" / "variance_ref").
Source Files